From practice changing to hypothesis generating: a virtuous cycle

This editorial refers to ‘Impact of renin–angiotensin system inhibitors on clinical outcomes in patients with severe aortic stenosis undergoing transcatheter aortic valve replacement: an analysis of from the PARTNER 2 trial and registries'†, by S. Chen et al., on page 943. Unlike legal matters where the level of evidence required to reach a verdict varies by jurisdictions, in medical decision-making the randomized controlled clinical trial (RCT) is uniformly recognized as the highest level of evidence across international borders. When an RCT generates robust data demonstrating the effectiveness of an intervention in altering the risks of fatal and serious non-fatal events, this evidence informs the interactions between caregivers and patients, and may reach the highest bar of improving medical practice. Randomization, the central pillar of the RCT, reduces the play of chance of confounding differences between the groups other than the treatment assignment. Thanks to the continued advances in surgical technique, post-operative care, and the absence of effective medical therapy, surgical aortic valve replacement (SAVR) was for long the definitive treatment of severe aortic stenosis (AS). On the basis of a series of high-quality RCTs showing comparative safety and efficacy (and even superiority in some populations) in regards to clinical outcomes and quality of life of transcatheter aortic valve replacement (TAVR) compared with the historical standard of care of SAVR, TAVR emerged as a viable, indeed preferred, option for a large subset of patients with severe symptomatic AS.1,2 As critical evidence from randomized trials has made TAVR a new therapeutic option in the treatment of AS, this frameshift has been acknowledged in current guidelines as well as in the recent Food and Drug Administration (FDA) approval of TAVR in patients at low surgical risk.3,4 Through expansion of TAVR into lower risk populations, enhancements in transcatheter valve platforms and system deployment, and growing operator proficiency outcomes following TAVR have continuously improved and remain targets for further fine-tuning. The surgical and interventional cardiology communities, joining forces in these landmark studies as ‘Heart Teams’, should be commended in executing these difficult studies against the gold standard of surgery. Continued questions regarding bioprosthetic valve durability and repeat valve-in-valve intervention will need to be resolved, which will require extended follow-up, and will greatly impact risk–benefit deliberations among patients and caregivers. Beyond these major mechanical considerations, other adjunctive therapies to minimize residual risk following aortic valve replacement in this generally elderly patient population with significant comorbidities are being considered. Despite the expanding skillset and expertise of the interventional cardiologist, and indeed of the Heart Team, it is the general cardiologist who first encounters the patient with AS in clinical practice and who, based on history and physical exam including old fashioned palpation and auscultation, establishes the diagnosis. Echocardiography is then used to confirm the clinical diagnosis, grade the severity of valvular stenosis, evaluate for other valvular lesions, and characterize left ventricular (LV) structure and function. The treatment of severe symptomatic AS remains interventional with aortic valve replacement be it SAVR or TAVR, as no medical therapy has been shown to favourably alter the disease course or prognosis. In this issue of the European Heart Journal, Chen et al. investigate the association between baseline use of either angiotensin-converting enzyme inhibitors (ACEIs) or angiotensin II receptor blockers (ARBs) on mortality in a large cohort of patients who underwent TAVR within the PARTNER 2 trials and registries.5 Among the 3979 TAVR recipients, whose operative risk ranged from intermediate to prohibitive, they identified 1736 with and 2243 without ACEI/ARB use prior to their TAVR. The unadjusted 2-year mortality was strikingly lower in those clinically selected for ACEI/ARB treatment [18.6% vs. 27.5%; absolute difference 8.9% in 2-year mortality; hazard ratio (HR) 0.64, 95% confidence interval (CI) 0.56–0.73; P < 0.001]. Many of the differences in mortality emerged after the initial 30 days following TAVR as unadjusted 30-day mortality post-TAVR did not differ significantly between the two groups (2.3% vs. 3.2%; P = 0.10). Considering the advanced age of the cohort (mean age of 82 years) this near 9% absolute increase in survival at 2 years, if true, is indeed remarkable. However, as anticipated from any non-randomized creation of a subgroup, there are numerous measured as well as unmeasured differences in baseline characteristics between those clinically selected for ACEI/ARB treatment and those who were not. To minimize these differences, the investigators used multivariable adjusted analyses as well as propensity score matching in attempts to reduce effects of measured confounding factors. In the multivariable adjusted analysis, stratified by study, 2-year mortality still remained significantly and strikingly lower in TAVR recipients treated with ACEIs/ARBs compared with untreated patients (HR 0.70, 95% CI 0.60–0.82; P < 0.001). In an important next step, the investigators used propensity matching, which reduced the sample size to 2520 patients (63% of the initial cohort). Once again, in the propensity score-matched cohort analysis, 2-year mortality remained significantly and impressively lower in patients with, compared with without, baseline ACEI/ARB use (HR 0.75, 95% CI 0.63–0.90; P = 0.001). The investigators should be commended for their well-conducted observational study, which expands on three recent retrospective cohort studies from three independent registries suggesting mortality benefit with post-TAVR renin–angiotensin system (RAS) inhibition. In the OCEAN-TAVI registry, ACEI/ARB use following TAVR was associated with a trend towards improved 2-year survival, which did not reach statistical significance after propensity matching (5.2% vs. 14.3%, P = 0.055).6 In contrast, larger observational studies from the RASTAVI and TVT registries showed significant reductions in cardiovascular (3-year cardiovascular mortality 7.2% vs. 11.3%, P = 0.006)7 and in all-cause mortality (1-year mortality 12.5% vs. 14.9%; HR 0.82, 95% CI 0.76–0.90) with prescription of RAS inhibitors upon hospital discharge.8 Collectively, these studies support the hypothesis that use of a RAS inhibitor may be beneficial for those undergoing TAVR. Despite cogent references to a biological plausibility for this striking observation such as reduction in wall stress and attenuation of LV hypertrophy and fibrosis, the mechanisms accounting for the observed survival benefit are not so clear. Importantly ACEI/ARB use has not been shown to favourably alter the prognosis of AS in (small) RCTs.9–11 Yet the magnitude and consistency of the observed survival differences with RAS inhibition after TAVR noted in three independent registries make the observation worthy of further examination—the focus of an RCT. The strength of the current study by Chen et al. lies in the uniqueness of the population, meticulous data collection with extensive information on possible confounding factors including frailty, echocardiographic review by an independent core laboratory, adjudication for cardiovascular and non-cardiovascular death, large sample size, and length of follow-up. In contrast to previous studies, only a small percentage of patients were excluded from the analysis. The investigators succinctly and appropriately enumerate the limitations of their observational study. As with any observational study and as the investigators appropriately state, inherent selection bias, residual confounding, and confounding by indication have to inform interpretation of these findings, which need to be viewed as hypothesis generating. In addition, the magnitude of the estimated effect of RAS inhibition on the reduction in cardiovascular and all-cause mortality in these elderly TAVR recipients, despite a nearly 20% cross-over by 2 years, appears greater than what could be anticipated by the administration of any currently known drug, and cries out for a follow-up focused RCT on this intervention. Additional data that would have enhanced this study and should be obtained prior to launching a large RCT would be the relative safety of ACEI/ARB treatment in this fragile population including altering rates of hospitalizations, worsening renal function, and hyperkalaemia. Despite these impressive findings from Chen et al.5 and three other observational studies, the absence of randomization precludes the recommendation of routine ACEI/ARB use in TAVR recipients, but raises a hypothesis that must be tested prospectively. Fortunately, this question is being addressed in the ongoing RASTAVI trial, which is assessing clinical outcomes and LV remodelling in 336 TAVR recipients randomized to ramipril vs. standard therapy post-TAVR, with an anticipated follow-up of 3 years.12 This study is designed with a most optimistic treatment effect (anticipated relative risk reduction of 0.55 for all-cause death) and therefore the limited sample size may render it underpowered to detect more conservative though clinically relevant differences in its primary endpoint of mortality, heart failure hospitalization, and stroke. Regardless of RAS inhibitor use, the well-conducted RCTs of TAVR for the treatment of severe AS have provided the evidence to dramatically change clinical practice, inform guidelines, improve public health, and expand the dialogue between patients and their caregivers (Figure 1, cycle A). The careful scrutiny of the trial and registry data by Chen and colleagues allowed the investigators to generate a new hypothesis to be tested about the use of RAS inhibition (Figure 1, cycle B), raise important questions, and make meaningful observations, which will in turn inform future RCTs and will stimulate basic investigation to better elucidate the underlying mechanisms (Figure 1, cycle C). RCTs provide the highest level of evidence in medicine and inform clinical guidelines if recommendations are supported by robust clinical outcomes. Quality indicators and performance measures improve quality of healthcare and thereby public health. (Cycle A) New observations derived from data from these studies generate new hypotheses to be tested in subsequent RCTs (cycle B), and stimulate basic research (cycle C), thus refuelling the virtuous cycle. RCTs provide the highest level of evidence in medicine and inform clinical guidelines if recommendations are supported by robust clinical outcomes. Quality indicators and performance measures improve quality of healthcare and thereby public health. (Cycle A) New observations derived from data from these studies generate new hypotheses to be tested in subsequent RCTs (cycle B), and stimulate basic research (cycle C), thus refuelling the virtuous cycle. Decades ago, in a similar virtuous cycle, fewer myocardial infarctions were first observed in patients randomized to ACEIs compared with placebo in both Survival and Ventricular Enlargement (SAVE) and Studies of Left Ventricular Dysfunction (SOLVD).13,14 Although of great interest, these non-primary outcomes were not sufficient to change clinical practice and were considered hypothesis generating. The question raised appropriately became the focus of specifically designed RCTs, which showed a reduction in cardiovascular events and death with use of ACEIs in a previously unstudied population.15 In addition to changing clinical practice, these findings stimulated further fundamental research to uncover important pathways between RAS inhibition and atherosclerosis, underscoring how in medicine secondary analyses of RCTs can generate hypothesis-generating information to be pursued more vigorously in RCTs (a non-vicious, indeed virtuous cycle, Figure 1, cycles A, B, and C).16 Unlike legal trials, which have a final verdict, a clinical trial should be viewed as an open-ended quest for knowledge in a virtuous cycle. Conflict of interest: M.A.P. receives research support from Novartis. He serves as a consultant for AstraZeneca, DalCor, GlaxoSmithKline, Novartis, Novo Nordisk, Pfizer, Roche, and Sanofi. Other: stock options for DalCor. K.S.J. has no conflicts to declare. The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology.