Treatments for pulmonary arterial hypertension: navigating through a network of choices

This editorial refers to ‘Pulmonary arterial hypertension treatment: an individual participant data network meta-analysis’, by J. Moutchia et al., https://doi.org/10.1093/eurheartj/ehae049. Pulmonary arterial hypertension (PAH) is a serious and progressive disease characterized by pulmonary vascular remodelling resulting in chronic elevation in mean pulmonary artery pressure and pulmonary vascular resistance, leading to dyspnoea, exercise limitation, right ventricular failure, and premature death.1 PAH remains incurable and portends a poor prognosis, although survival rates have improved over the last decades due to the development of multiple medications and specialized care.2 The treatment landscape for PAH has evolved dramatically over the past 30 years.3 Medications targeting the endothelin, nitric oxide, and prostacyclin pathways have been developed: endothelin receptor antagonists, phosphodiesterase type 5 inhibitors, guanylate cyclase stimulators, prostacyclin derivatives, and prostacyclin receptor agonists.4 These treatments have shown clinical benefits by and large compared with placebo, including improved symptoms, functional class, exercise capacity, and reduced clinical worsening.2,5,6 However, the comparative analysis of different drugs for PAH is challenging. The rarity of the disease makes patient recruitment for clinical trials difficult. Randomized controlled trials (RCTs) often include patients with varied backgrounds, ages, and PAH aetiologies.7 Most PAH RCTs were relatively short term (12–16 week double-blind placebo-controlled periods), leading to low event rates for some clinical outcomes, such as death or lung transplantation, which limits statistical power to estimate the effect of treatment on these outcomes.4 There are some epidemiological methods that may be used to help to address these concerns. A network meta-analysis (NMA) allows for comparative analysis of treatment effects across different studies and interventions. However, aggregate data NMAs are fraught with challenges in and of themselves, including comparing aggregate RCT data, which may not fully satisfy the statistical assumptions of the NMA [i.e. transitivity (similarity of patients across trials) and coherence (agreement of indirect and direct evidence)] and not account for significant heterogeneity across trials. Our Graphical Abstract summarizes the IPD NMA process, benefits and disadvantages. To address these concerns, Moutchia and colleagues, in their study published in this issue of the European Heart Journal, performed an individual participant data (IPD) NMA, which gathered data from 20 RCTs submitted to the Food and Drug Administration (FDA), including 6811 PAH patients.8 As compared with an aggregate data NMA that uses data from RCTs, IPD meta-analysis leverages patient-level data from RCTs and re-analyses them, accounting for heterogeneity between trials using appropriate statistical methods such as hierarchical modelling. The study authors focused on the three classical PAH treatment pathways. Their primary outcomes included changes in 6-min walk distance (6MWD) and time to first clinical worsening. Secondary outcomes encompassed overall survival and haemodynamic parameters. They found that combined therapy targeting both the endothelin and nitric oxide pathways was effective at improving important patient outcomes compared with alternatives. The study was statistically rigorous and an impressive presentation of patient-level data. One of the major advantages and novelties of this IPD NMA was the ability to evaluate heterogeneity of treatment effects according to a variety of important comorbidities, which is a controversial topic in the PAH literature. Post-hoc analyses of treatment effects in specific trials have led to divergent findings about whether comorbidities modify treatment effects. The authors were able to evaluate effect modification by comorbidities across each treatment pathway, with some interesting results. For example, treatment effects on 6MWD decreased with older age with most pathways, except for intravenous/subcutaneous prostacyclin which appeared more effective in older patients, somewhat unexpectedly. Treatment effects on clinical worsening also tended to increase with higher body mass index, except for the nitric oxide pathway. Nitric oxide pathway therapies were less effective with higher body mass index and in the presence of hypertension, diabetes, or coronary artery disease. This introduces important new hypotheses to explore the personalization of therapeutic choices according to specific comorbidities. This study allows us to reflect on a few critical points regarding evidence synthesis in PAH. The first concerns the role of IPD meta-analysis and NMA as compared with aggregate data comparisons. A proposed strength of IPD meta-analysis is that it allows for more nuanced assessment of heterogeneity by removing inconsistencies in data analysis, subgroup analysis assessments, and adjustment for important co-variates across studies. Important limitations include resource intensiveness and risk of publication bias by including select trials reviewed by the FDA which probably represent positive trials and excludes trials with negative results. Furthermore, the study does not include informative trials such as more recent trials that directly compared triple vs. dual oral therapy.9 One important question that this study raises is if we need IPD analysis in a disease such as PAH. For example, existing aggregate data NMAs have essentially concluded the same findings as the present study and can assess the data using rigorous quality assessment procedures such as GRADE, which are fundamental to guideline development and recommendations.4 Modern PAH studies are relatively standardized and comparatively high quality, lending credibility and uniformity to aggregate data. Given the resource intensity, cost, and complexity associated with IPD, aggregate meta-analyses emerge as a more efficient and timely method. Furthermore, IPD meta-analyses have significant methodological pitfalls, which makes their interpretation and use in evidence synthesis uncertain.10 For example, although the present study has been completed with significant statistical rigour, the authors do not present their results using the GRADE method. Although GRADE has not been regularly implemented in IPD NMAs, without a quality assessment readers are left with an incomplete perspective on comparative efficacy. In addition, aggregate data NMA may use tools such as ICEMAN (i.e. a validated tool) to assess the credibility of subgroup effects, to provide confidence in a particular subgroup. This analysis is missing from the present study and is something to consider for future endeavours. Furthermore, publication bias is a significant concern. The present study was able to include 20 RCTs submitted to the FDA which excluded many negative studies, for example, as well as the large body of phase II trials. This is a major concern that study authors will need to address going forward. The landscape of PAH treatment has drastically changed over the past few decades and continues to develop at an accelerated pace. Breaking through the classical pathways, recent phase II and III trials showed that sotatercept is effective at reducing clinical worsening and improving 6MWD as compared with placebo, with most patients (>50%) in both arms on triple therapy treatment targeting the three classical pathways.11,12 Sotatercept is an activin signalling inhibitor that targets the transforming growth factor-β superfamily.13,14 Indeed, head-to-head trials comparing sotatercept with conventional PAH therapies may be difficult to execute, and what is greatly needed in PAH is advancement in evidence synthesis, such as provided by Moutchia and colleagues in their analysis. The authors proposed to make their review ‘living’—a concept that has become prominent since the COVID-19 pandemic. This may address some of the biases and limitations if, for example, phase II trials and negative trials can be included eventually and future therapies such as sotatercept are included. Future studies will need to address the comparative effectiveness of sotatercept as compared with conventional therapy to help guide clinicians and patients forward.15 J.W. discloses grants from Janssen, Bayer, Merck, and Astra Zeneca, and consulting fees from Janssen and Merck. He has received payments or honoraria for lectures/presentations and payments for expert testimony for sprigings intellectual property law. He has received support for attending meetings from Janssen (travel support). He also disclosed payments from Janssen, Merck, and Universite Laval for participation on a Data Safety Monitoring Board or advisory board. He holds a leadership/fiduciary role with the Pulmonary Hypertension Association of Canada. M.H. discloses grants or contracts from Acceleron, AOP orphan, Janssen, Merck, and Shou Ti. He has also received consulting feeds from 35 Pharma, Aerovate, AOP orphan, Bayer, Chiesi, Ferrer, Janssen, Kerros, Merck, MorphogenIX, Shou Ti, and United Therapeutics. He has received payments or honoraria for lectures/presentations from Janssen and Merck, and has participated on Data Safety Monitoring Board or Advisory Board for Accleron, Altavant, Janssen, Merck, and United Therapeutics. T.P. declares no disclosure of interest.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.