Endothelin receptor antagonists for pulmonary arterial hypertension

Abstract

Background

Pulmonary arterial hypertension is a devastating disease that leads to right heart failure and premature death. Endothelin receptor antagonists have shown efficacy in the treatment of pulmonary arterial hypertension.

Objectives

To evaluate the efficacy of endothelin receptor antagonists (ERAs) in pulmonary arterial hypertension.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and the reference sections of retrieved articles. The searches are current as of 4 November 2020.

Selection criteria

We included randomised trials and quasi‐randomised trials involving participants with pulmonary arterial hypertension.

Data collection and analysis

Two of five review authors selected studies, extracted data and assessed study quality according to established criteria. We used standard methods expected by Cochrane. The primary outcomes were exercise capacity (six‐minute walk distance, 6MWD), World Health Organization (WHO) or New York Heart Association (NYHA) functional class, Borg dyspnoea scores and dyspnoea‐fatigue ratings, and mortality.

Main results

We included 17 randomised controlled trials involving a total of 3322 participants. Most trials were of relatively short duration (12 weeks to six months). Sixteen trials were placebo‐controlled, and of these nine investigated a non‐selective ERA and seven a selective ERA.  

We evaluated two comparisons in the review: ERA versus placebo and ERA versus phosphodiesterase type 5 (PDE5) inhibitor. The abstract focuses on the placebo‐controlled trials only and presents the pooled results of selective and non‐selective ERAs.

After treatment, participants receiving ERAs could probably walk on average 25.06 m (95% confidence interval (CI) 17.13 to 32.99 m; 2739 participants; 14 studies; I² = 34%, moderate‐certainty evidence) further than those receiving placebo in a 6MWD. Endothelin receptor antagonists probably improved more participants' WHO functional class (odds ratio (OR) 1.41, 95% CI 1.16 to 1.70; participants = 3060; studies = 15; I² = 5%, moderate‐certainty evidence) and probably lowered the odds of functional class deterioration (OR 0.43, 95% CI 0.26 to 0.72; participants = 2347; studies = 13; I² = 40%, moderate‐certainty evidence) compared with placebo. There may be a reduction in mortality with ERAs (OR 0.78, 95% CI 0.58, 1.07; 2889 participants; 12 studies; I² = 0%, low‐certainty evidence), and pooled data suggest that ERAs probably improve cardiopulmonary haemodynamics and may reduce Borg dyspnoea score in symptomatic patients. Hepatic toxicity was not common, but may be increased by ERA treatment from 37 to 67 (95% CI 34 to 130) per 1000 over 25 weeks of treatment (OR 1.88, 95% CI 0.91 to 3.90; moderate‐certainty evidence). Although ERAs were well tolerated in this population, several cases of irreversible liver failure caused by sitaxsentan have been reported, which led the licence holder for sitaxsentan to withdraw the product from all markets worldwide. 

As planned, we performed subgroup analyses comparing selective and non‐selective ERAs, and with the exception of mean pulmonary artery pressure, did not detect any clear subgroup differences for any outcome.

Authors' conclusions

For people with pulmonary arterial hypertension with WHO functional class II and III, endothelin receptor antagonists probably increase exercise capacity, improve WHO functional class, prevent WHO functional class deterioration, result in favourable changes in cardiopulmonary haemodynamic variables compared with placebo. However, they are less effective in reducing dyspnoea and mortality. The efficacy data were strongest in those with idiopathic pulmonary hypertension. The irreversible liver failure caused by sitaxsentan and its withdrawal from global markets emphasise the importance of hepatic monitoring in people treated with ERAs. The question of the effects of ERAs on pulmonary arterial hypertension has now likely been answered. The combined use of ERAs and phosphodiesterase inhibitors may provide more benefit in pulmonary arterial hypertension; however, this needs to be confirmed in future studies.

Plain language summary

Endothelin receptor antagonists for pulmonary arterial hypertension

Review question

Do endothelin receptor antagonists increase how much a person is capable of exercising (exercise capacity), improve symptoms, or reduce death in people with pulmonary arterial hypertension (PAH)?

Background

Pulmonary arterial hypertension is a devastating disease characterised by an increase in pulmonary vascular resistance which leads to right heart failure and ultimately death.

Endothelin receptor antagonists are a class of strong vasodilators (medications that open (dilate) blood vessels) capable of stopping the process of cell division, which could dilate and result in a favourable pulmonary arterial structural alteration.

Study characteristics

We reviewed the evidence from randomised studies (studies in which people are assigned to one of two or more treatment groups using a random method). After a thorough search and assessment of the medical literature, we identified 17 studies with a total of 3322 participants for inclusion in the review. A vast majority of the participants had PAH without known cause (idiopathic). The evidence is current to November 2020.

Key results

Endothelin receptor antagonists probably increase exercise capacity, improve World Health Organization functional class (a measurement of how severe a person's pulmonary hypertension symptoms are), and may improve death rates and symptoms in people with PAH; however they may also increase the risk of liver damage, although this was rare. The question of the effects of endothelin receptor antagonists on PAH has now likely been answered.

Certainty of the evidence

Overall, the evidence presented is of moderate certainty due to the high occurrence of missing data.

Summary of findings

Summary of findings 1. Endothelin receptor antagonists compared to placebo for pulmonary arterial hypertension.

Endothelin receptor antagonists compared to placebo for pulmonary arterial hypertension
Participant or population: pulmonary arterial hypertension Setting: clinics and hospitals Intervention: endothelin receptor antagonists Comparison: placebo
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo	Risk with endothelin receptor antagonists
Change from baseline in 6MWD (m) mean duration of study 16.3 weeks	The weighted mean change on control was −4.56 m.	MD 25.06 higher (17.13 higher to 32.99 higher)	‐	2739 (14 RCTs)	⊕⊕⊕⊝ Moderate1	Higher is better for 6MWD.
Proportion of participants with improved functional class mean duration of study 16.8 weeks	175 per 1000	230 per 1000 (197 to 264)	OR 1.41 (1.16 to 1.70)	3060 (15 RCTs)	⊕⊕⊕⊝ Moderate1	Participants with high OR are more likely to achieve functional improvement.
Change from baseline in BDI mean duration of study 14.3 weeks	The weighted mean change on control was 0.25 higher.	MD 0.43 lower (0.90 lower to 0.04 higher)	‐	788 (7 RCTs)	⊕⊕⊝⊝Low2	Symptoms are worse with higher score of BDI.
Mortality mean duration of study 30.2 weeks	73 per 1000	58 per 1000 (44 to 78)	OR 0.78 (0.58 to 1.07)	2889 (12 RCTs)	⊕⊕⊝⊝ Low3	Participants with lower OR are less likely to die.
Change from baseline in mean PAP (mmHg) mean duration of study 17.1 weeks	The weighted mean change on control was 0.53 higher.	MD 4.65 lower (6.05 lower to 3.26 lower)	‐	729 (8 RCTs)	⊕⊕⊕⊝ Moderate4	Participants are worse with higher pulmonary artery pressure.
Change from baseline in PVR (dyn/s/cm5) mean duration of study 15.7 weeks	The weighted mean change on control was 63.55 higher.	MD 236.24 lower (333.21 lower to 139.26 lower)	‐	586 (7 RCTs)	⊕⊕⊕⊝ Moderate4	Participants are worse with higher pulmonary vascular resistance.
Hepatic toxicity mean duration of study 25 weeks	37 per 1000	67 per 1000 (34 to 130)	OR 1.88 (0.91 to 3.90)	2250 (11 RCTs)	⊕⊕⊕⊝ Moderate1	Participants with higher OR are more likely to suffer hepatic toxicity.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). 6MWD: 6‐minute walk distance; BDI: Borg dyspnoea index; CI: confidence interval; MD: mean difference; OR: odds ratio; PAP: pulmonary artery pressure; PVR: pulmonary vascular resistance; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

¹Incomplete outcome data (attrition bias) due to missing data imbalanced between intervention and control groups in most of the included studies (−1 level).
²Attrition, publication bias and wide upper confidence interval (−2 levels).
³We downgraded the evidence for the outcome of mortality by 2 levels because the study period was relatively short and events in most of the studies few, and wide upper confidence interval of odds ratio.
⁴The pooled analysis used raw data from subgroups of the included studies.

Background

Description of the condition

Pulmonary arterial hypertension (PAH) is a rare disease of the distal pulmonary arteries characterised by vasoconstriction, vascular proliferation, obstructive remodelling of the pulmonary vessel wall, inflammation, and thrombosis, which lead to a progressive increase in pulmonary vascular resistance, and, ultimately, right ventricular failure and death (Galiè 2015; Thenappan 2018). PAH includes the following subcategories (Simonneau 2019): 1. idiopathic PAH; 2. heritable PAH; 3 drug‐ and toxin‐induced PAH; 4. PAH associated with (4.1 connective tissue disease; 4.2 HIV infection; 4.3 portal hypertension; 4.4 congenital heart disease; 4.5 schistosomiasis); 5. PAH long‐term responders to calcium channel blockers; 6. PAH with overt features of venous/capillaries involvement; and 7. persistent pulmonary hypertension of the newborn (PPHN) syndrome.

Considerable progress has been made in identifying the causes of PAH in recent decades. Mutations in the type II bone morphogenetic protein receptor (BMPR2) gene contributes to heritable PAH (Thenappan 2018). Epigenetic dysregulation of DNA methylation, histone acetylation, and microRNAs play a key role in disease pathogenesis (Galiè 2015). Complex changes in cytokines, cellular immunity, and autoantibodies indicate that PAH is in part an autoimmune, inflammatory disease (Guignabert 2013). Endothelial dysfunction characterised by overexpression of vasoconstrictors and downexpression of vasoactive mediators occurs in PAH as a result. Reduced production of vasoactive mediators such as nitric oxide (NO) and prostacyclin, along with overexpression of vasoconstrictors such as endothelin‐1 (ET‐1) and thromboxane A₂, increase pulmonary vascular tone and promote pulmonary vascular remodelling in people with PAH. The substances targeting pulmonary vasoactive mechanisms (Galiè 2015; Klinger 2019; Thenappan 2018), used alone or in combination, therefore improve functional capacity and haemodynamics and reduce hospital admissions.

Description of the intervention

Endothelin receptor antagonists (ERAs) are a class of potent vasodilators and antimitotic substances that could specifically dilate and remodel the pulmonary arterial system (Horinouchi 2013). Two separate receptors for ET have been identified: endothelin type A (ET‐A) receptors, which are most commonly found on vascular smooth muscle cells and which induce vasoconstriction by increasing intracellular calcium; and endothelin type B (ET‐B) receptors located on endothelial cells, which stimulate the release of vasodilating agents such as NO and prostacyclin (Madonna 2015). However, ET‐B receptors also appear on vascular smooth muscle cells, where they stimulate vasoconstriction. Theoretically, either selective blocking of ET‐A receptors alone, or non‐selective blocking of both ET‐A and ET‐B receptors together, could dilate local pulmonary vessels (Klinger 2019; Thenappan 2018).

How the intervention might work

Endothelin is a potent vasoconstrictor and smooth muscle mitogen. It is overexpressed in the lungs of people with PAH, and elevated plasma endothelin concentrations are correlated with poorer prognosis (Chester 2014). Furthermore, an overproduction of endothelin may also lead to pulmonary vascular remodelling in PAH (Shao 2011). ERAs have shown efficacy in the treatment of PAH in clinical trials (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; COMPASS‐2; SERAPHIN; STRIDE‐1; STRIDE‐2).

Why it is important to do this review

Since bosentan, the first ERA, was approved by the US Food and Drug Administration (FDA) to treat PAH, many new agents, both selective and non‐selective ERAs, have been developed and tested in clinical trials. Importantly, the efficacy and safety of combined use of ERAs and phosphodiesterase (PDE) inhibitors have been investigated in population with PAH (AMBITION; COMPASS‐2). However, the efficacy of ERAs was not consistent in these clinical trials; a systematic review to determine the efficacy of ERAs in the treatment of PAH was thus warranted.

Objectives

To evaluate the efficacy of endothelin receptor antagonists in pulmonary arterial hypertension.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials or quasi‐randomised clinical trials.

Types of participants

We included trials involving adults and children (≥ 2 years) with PAH. The diagnosis of PAH should have been made according to European Respiratory Society/European Society of Cardiology/World Health Organization (ERS/ESC/WHO) guidelines. We included studies involving participants with PAH secondary to other diseases if data assessing the outcomes specific to the PAH component of their syndrome were available.

Types of interventions

We considered trials in which participants took an ERA alone or in combination against any comparator.

We included the following co‐interventions.

1. Phosphodiesterase type 5 (PDE5) inhibitors

2. Soluble guanylate cyclase (sGC)

3. Prostanoids

4. Nitrates

5. Calcium channel blockers

6. Non‐pulmonary hypertension‐specific medications including diuretics, anticoagulants, and oxygen

Types of outcome measures

Primary outcomes

1. Exercise capacity (as measured by a six‐minute walk distance (6MWD)).

2. World Health Organization (WHO) functional class or New York Heart Association (NYHA) functional class (WHO/NYHA).

3. Borg dyspnoea scores and dyspnoea‐fatigue ratings.

4. Mortality.

Secondary outcomes

1. Cardiopulmonary haemodynamics including mean pulmonary artery pressure (PAP); pulmonary vascular resistance (PVR), cardiac index, cardiac output (CO), systemic arterial oxygen saturation and systemic oxygen transport.

2. Pulmonary function tests.

3. Adverse events (e.g. hepatic toxicity).

Search methods for identification of studies

Electronic searches

The Cochrane Airways Group Information Specialist undertook literature searches on 4 November 2020 using the following sources: the Cochrane Central Register of Controlled Trials (CENTRAL, Issue 11 of 12, November 2020) (from 1966 to November 2020), which contains the Cochrane Airways Group Register of Trials, MEDLINE Ovid SP (from 1966 to November 2020), Embase Ovid SP (from 1980 to November 2020), the US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov/), and the World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch/). We applied no language restrictions. The database search strategies are shown in Appendix 1.

Searching other resources

We handsearched the reference lists of retrieved articles for additional trials. We also searched documents released from the FDA and European Medicines Agency (EMA).

Data collection and analysis

Selection of studies

Three review authors (LC, CJ, and DB) independently screened the titles and abstracts of records identified by the search for potential eligibility, categorising them as 'retrieve' or 'do not retrieve'. We retrieved all available full‐text publications. Three review authors (LC, CJ, and DB) independently assessed the full‐text reports as either eligible for inclusion, or identified and reported the reasons for exclusion (Figure 1). Any disagreements were resolved by consensus or by involving a third person/review author (GY). One final review author (LK) analysed the included and excluded texts to ensure uniform enforcement of the study protocol. We excluded duplicates and collated multiple reports of the same study (Figure 1). We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Figure 1) and 'Characteristics of excluded studies' table.

1.

Study flow diagram.

Data extraction and management

We used Covidence to collect study characteristics and outcome data (Covidence). Three review authors (LC, CJ, and GY) extracted the following study characteristics from the included studies.

1. Methods: study design, total duration of study, details of any 'run‐in' period, number of study centres and location, study setting, withdrawals, and date of study.

2. Participants: N, mean age, age range, gender, severity of condition, diagnostic criteria, baseline lung function, smoking history, inclusion criteria, and exclusion criteria.

3. Interventions: intervention, comparison, concomitant medications, and excluded medications.

4. Outcomes: primary and secondary outcomes specified and collected, and time points reported.

5. Notes: funding for studies and notable conflicts of interest of trial authors.

Three review authors (LC, CJ, and GY) independently extracted the outcome data from included studies. Any disagreements were resolved through discussion or by consulting a third review author (DB) if required. One review author (LC) transferred data into the Review Manager 5 file (Review Manager 2014). We double‐checked that data were entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (LK) spot‐checked study characteristics for accuracy against the study report.

Assessment of risk of bias in included studies

We independently assessed the methodological quality of included trials against the following criteria.

1. Was the study described as randomised?

2. Was the allocation concealment adequate?

3. Were the participants blinded?

4. Was there blinded outcome assessment?

5. Was there a description of withdrawals and dropouts?

6. Were the results analysed according to intention‐to‐treat?

7. Was there any other bias?

We graded each potential source of bias as 'low', 'high', or 'unclear', and provided a justification for our judgement in the 'Risk of bias' table. We took into account the risk of bias when considering treatment effects.

Measures of treatment effect

We treated short ordinal scales (e.g. WHO functional class, NYHA class) as dichotomous measures, arranging the data into two categories: improvement or deterioration in class. We analysed linear scales (e.g. Borg dyspnoea scores) as continuous data.

Unit of analysis issues

For studies where more than one eligible treatment arm was included, we split the data between two treatment‐control comparisons by splitting the control group and halving the sample sizes. The total numbers in the meta‐analysis thus added up to the original size of the group.

Dealing with missing data

For missing data on prespecified primary outcomes, we contacted the study investigators to request the data. If the data were not available, we performed sensitivity analyses to assess how sensitive the results were to reasonable changes in the assumptions made.

Assessment of heterogeneity

We compared the Chi² statistic with its degrees of freedom to identify heterogeneity. We considered a statistic bigger than its degrees of freedom and I² > 30% as evidence of heterogeneity.

Assessment of reporting biases

We conducted standard funnel plots for the primary outcome (i.e. 6MWD) to investigate the potential for the influence of publication bias on the analysis.

Data synthesis

In the absence of significant heterogeneity, we combined results using a mean difference (MD) for continuous data, and odds ratio (OR) for dichotomous data. If there was significant heterogeneity (i.e. the ratio of Chi² statistic and its degrees of freedom was bigger than 1.0, and the I² > 30%), we used the random‐effects model to assess treatment effect.

A number of articles reported studies where more than one eligible treatment arm was included. For these studies, we split the data between two treatment‐control comparisons by splitting the control group and halving the sample sizes. For example, STRIDE‐1 was a triple‐arm clinical study that did not provide combined 'sitaxsentan' data versus placebo on secondary outcomes such as mean pulmonary artery pressure (mPAP) and pulmonary vascular resistance (PVR). For relevant outcomes, we entered data from both intervention arms separately for the meta‐analysis, denoting the dosage distinction in footnotes. Data for placebo participants were entered with the number of participants in each of the placebo 'groups' halved. The total numbers in the meta‐analysis thus added up to the original size of the group. The same method was used in STRIDE‐2, STRIDE‐4, and SERAPHIN (see Characteristics of included studies). STRIDE‐2 compared sitaxsentan with both a placebo arm and an open‐label bosentan arm. For this study, we used the data from participants treated with bosentan and placebo for reference in the notes, and did not pool the data with those of other studies.

The AMBITION study had triple arms, that is combination therapy with 10 mg of ambrisentan plus 40 mg of tadalafil (combination‐therapy group), 10 mg of ambrisentan plus placebo (ambrisentan monotherapy group), or 40 mg of tadalafil plus placebo (tadalafil‐monotherapy group). We pooled the data from the combination‐therapy groups versus tadalafil‐monotherapy group in the review, and compared ambrisentan monotherapy versus tadalafil in the subgroup analysis of ERAs versus phosphodiesterase inhibitors.

Subgroup analysis and investigation of heterogeneity

1. Non‐selective ERA (antagonising both ET‐A and ET‐B) versus selective ERA (antagonising ET‐A) only.

2. Participants with idiopathic PAH versus those with PAH related to other conditions.

Sensitivity analysis

When there was a significant difference in the characteristics of participants, intervention, comparator, or outcome, we conducted a sensitivity analysis to assess how the results were affected by reasonable changes to the assumptions made.

Summary of findings and assessment of the certainty of the evidence

We created a 'Summary of findings' table using the following outcomes (Table 1): change in 6MWD, improvement in WHO functional class, Borg dyspnoea scores, mortality, PAP, PVR, and risk of hepatic toxicity. We used the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence as it relates to the studies that contribute data for the prespecified outcomes. We used the methods and recommendations described in Section 8.5 and Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020), employing GRADEpro GDT software (GRADEpro GDT). We justified all decisions to downgrade the quality of studies using footnotes and made comments to aid the reader's understanding of the review where necessary.

Results

Description of studies

Results of the search

The literature search identified 5154 titles and abstracts. Of these, we identified 17 randomised controlled trials for inclusion in this version of the review (Table 2) (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; COMPASS‐2; EARLY; EDITA; MAESTRO; PORTICO; SERAPH; SERAPHIN; STRIDE‐1; STRIDE‐2; STRIDE‐4). We divided four trials based on the doses given (SERAPHIN; STRIDE‐1; STRIDE‐2; STRIDE‐4); see Data synthesis for details. The process of identifying studies eligible for inclusion in the analysis is summarised in a flow diagram (Figure 1). We obtained the full‐text versions of the 17 included studies from published papers or the FDA and EMA websites. Galiè 2003 reported data from a subset of participants from the BREATHE‐1 study. Studies referenced in multiple sources were assessed for completeness of study details. When there was a discrepancy in the data between published journal papers and documents from the FDA and EMA, we used data from the FDA or EMA. We identified two potentially eligible studies in abstract form. We contacted the authors for detailed data, but received no response, therefore we assessed these studies as awaiting classification (for details see Characteristics of studies awaiting classification).

1. Main study characteristics across all studies.

Study	N	Country	Intervention	Control	Outcomes
AMBITION	747	International	Ambrisentan + tadalafil or ambrisentan	Tadalafil + placebo	Primary outcome: time to the first event of clinical failure Secondary outcomes: change from baseline in NT‐proBNP level, 6MWD, WHO FC, and Borg dyspnoea index
ARIES‐1	201	International	Ambrisentan (5 mg/day or 10 mg/day)	Placebo	Primary outcome: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening, plasma BNP, and SF‐36 physical functioning scale
ARIES‐2	192	International	Ambrisentan (2.5 mg/day or 5 mg/day)	Placebo	Primary outcome: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening, plasma BNP, and SF‐36
BREATHE‐1	213	International	Bosentan	Placebo	Primary outcome: change from baseline in 6MWD Secondary outcomes: Borg dyspnoea index, WHO FC
BREATHE‐2	33	International	Bosentan	Placebo	Primary outcome: change from baseline to week 16 in TPR Secondary outcomes: change in cardiac index, PVR, mPAP, mRAP, 6MWD, NYHA FC and dyspnoea‐fatigue rating
BREATHE‐5	54	International	Bosentan	Placebo	Primary outcome: change from baseline in SpO2 and PVR Secondary outcomes: change from baseline in 6MWD, WHO FC, cardiac index, PVR, mPAP, and mRAP
Channick 2001	32	International	Bosentan	Placebo	Primary outcome: change from baseline in 6MWD Secondary outcomes: cardiopulmonary haemodynamics (cardiac index, PVR, mPAP, mRAP), WHO FC
COMPASS‐2	334	International	Bosentan	Placebo	Primary outcome: time to the first morbidity/mortality event Secondary outcomes: change in 6MWD, WHO FC, time to the first occurrence of death from any cause, hospitalisation for PAH or start of intravenous prostanoid therapy, atrial septostomy, or lung transplant.
EARLY	185	International	Bosentan	Placebo	Primary outcomes: PVR and change from baseline in 6MWD Secondary outcomes: time to clinical worsening, change from baseline to month 6 in WHO FC, Borg dyspnoea index, mPAP, cardiac index, RAP, and SvO2
EDITA	38	Germany	Ambrisentan	Placebo	Primary outcome: change in mPAP Secondary outcomes: change in WHO FC, change in cardiac index, change in PVR, symptoms of SSc, quality of life (SF‐36), lung function tests, right heart dimensions and function, NT‐proBNP, measures of disease‐related progression
MAESTRO	150	International	Macitentan	Placebo	Primary outcome: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC and Borg dyspnoea index
PORTICO	85	International	Macitentan	Placebo	Primary outcome: change from baseline to PVR Secondary outcomes: change from baseline in RAP, mPAP, cardiac index, total pulmonary resistance, SvO2, NT‐proBNP, 6MWD, and WHO FC
SERAPH	26	British	Bosentan	Sildenafil	Primary outcome: change in right ventricle mass from baseline Secondary outcomes: change from baseline in 6MWD, cardiac index, Borg dyspnoea index, quality of life, and plasma BNP level from baseline
SERAPHIN	742	International	Macitentan	Placebo	Primary outcome: time from the initiation of treatment to the first event related to pulmonary arterial hypertension Secondary outcomes: change from baseline in 6MWD, percentage of participants with an improvement in WHO FC, death due to PAH or hospitalisations for PAH, and death from any cause
STRIDE‐1	178	International	Sitaxsentan	Placebo	Primary outcome: peak oxygen consumption Secondary outcomes: change from baseline in 6MWD, NYHA FC, PAP, cardiac index, and PVR
STRIDE‐2	245	International	Sitaxsentan	Placebo	Primary outcome: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening
STRIDE‐4	98	International	Sitaxsentan	Placebo	Primary efficacy endpoint was the change in 6MWD from baseline to week 18. Secondary outcomes: changes in WHO FC from baseline at each assessment and time to clinical worsening, Borg dyspnoea index

6MWD: 6‐minute walk distance; BNP: B‐type natriuretic peptide; mPAP: mean pulmonary artery pressure; mRAP: mean right atrial pressure; MVO₂: mixed venous oxygen saturation; NT‐proBNP: N‐terminal pro–brain natriuretic peptide; NYHA FC: New York Heart Association functional class; PAH: pulmonary arterial hypertension; PVR: pulmonary vascular resistance; SF‐36: 36‐item Short Form Health Survey; SpO₂: oxygen saturation; SSc: systemic sclerosis; SvO₂: mixed venous oxygen saturation; TPR: total pulmonary resistance;  WHO FC: World Health Organization functional class

Included studies

Methods

Sixteen studies were randomised and placebo‐controlled (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; COMPASS‐2; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1; STRIDE‐2; STRIDE‐4). One study was a randomised head‐to‐head study of bosentan (an ERA) and sildenafil (a PDE5 inhibitor) (SERAPH). A full description of each study with clinical characteristics at baseline, inclusion/exclusion criteria, methods of randomisation, and outcomes is provided in Characteristics of included studies.

Interventions

Nine trials compared a non‐selective ERA (bosentan and macitentan) against placebo (BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; COMPASS‐2; EARLY; MAESTRO; PORTICO; SERAPHIN). Seven trials compared selective ERAs (sitaxsentan or ambrisentan) against placebo (AMBITION; ARIES‐1; ARIES‐2; EDITA; STRIDE‐1; STRIDE‐2; STRIDE‐4). Three studies tested the efficacy of combination therapy versus monotherapy in participants with PAH (AMBITION; BREATHE‐2; COMPASS‐2). SERAPH compared bosentan against sildenafil (a PDE5 inhibitor).

For the purposes of the review, we specified two comparisons: comparison 1: ERA versus placebo (where ERAs are subgrouped into selective versus non‐selective); and comparison 2: ERA versus PDE5 inhibitor.

Participants

Thirteen trials recruited participants with idiopathic PAH and PAH secondary to connective tissue disease and congenital heart disease (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; Channick 2001; COMPASS‐2; EARLY; SERAPH; SERAPHIN; STRIDE‐1; STRIDE‐2; STRIDE‐4). Two trials tested the efficacy of ERAs in participants with Eisenmenger syndrome (BREATHE‐5; MAESTRO). One study investigated the effect of macitentan on participants with portopulmonary hypertension (PORTICO). One study investigated systemic sclerosis patients with mildly elevated mean pulmonary arterial pressure (mPAP) at rest between 21 and 24 mmHg and/or > 30 mmHg during low‐dose exercise (EDITA).

Outcomes

Seven trials used the 6MWD as the primary outcome measure (ARIES‐1; ARIES‐2; BREATHE‐1; Channick 2001; MAESTRO; STRIDE‐2; STRIDE‐4), whilst one trial used change from baseline in peripheral oxygen saturation (SpO₂) and PVR as the primary outcome (BREATHE‐5). STRIDE‐1 used peak exercise VO₂ as its primary outcome. EARLY and PORTICO used PVR as the primary outcome. The primary efficacy parameter in BREATHE‐2 was change from baseline to week 16 in total pulmonary resistance (TPR). SERAPH used change in right ventricle mass from baseline as its primary endpoint. Three studies used time to first event of clinical worsening as the primary endpoint (AMBITION; COMPASS‐2; SERAPHIN). The EDITA study used change in mPAP at rest between groups as the primary endpoint. Additional outcome measures used in the included studies were WHO or NYHA functional class status, Borg dyspnoea score, cardiopulmonary haemodynamics, and drug safety and tolerability. For details, see Characteristics of included studies.

Excluded studies

We excluded 101 studies, with reasons provided in the Characteristics of excluded studies table.

Risk of bias in included studies

All but two trials, EDITA; SERAPH, were reported as international, multicentre, double‐blind, randomised, placebo‐controlled trials. Both SERAPH and EDITA are single‐centre, double‐blind, randomised, placebo‐controlled trials. Randomisation in all trials was conducted according to computer‐generated random number programs. We judged allocation concealment to be adequate in all trials. All trials described dropouts and withdrawals. All but two trials, EARLY; EDITA, were statistically analysed according to the intention‐to‐treat principle. See Characteristics of included studies for a full description of the risk of bias in included studies, and Figure 2 for a 'Risk of bias' summary.

2.

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.

Allocation

All of the included trials used computer‐generated random numbers and were placebo controlled, therefore we judged allocation concealment to be adequate.

Blinding

Eight trials reported detailed descriptions of how treating clinicians, participants, and assessors were sufficiently blinded and were therefore judged as at low risk of bias (AMBITION; BREATHE‐5; COMPASS‐2; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN). Other trials were reported to be double‐blind, but as no further information was available we judged them to be at unclear risk of bias.

Incomplete outcome data

All studies described details of dropouts and withdrawals. All trials except for two, EARLY; EDITA, performed statistical analysis based on intention‐to‐treat. However, due to the high occurrence of missing data, 13 included trials were categorized as at high risk of attrition bias. For details, see Characteristics of included studies.

Selective reporting

Funnel plots for trials comparing ERAs versus placebo (outcome of change from baseline in 6MWD) were qualitatively symmetrical and did not suggest the presence of publication bias. Nine trials reported their prespecified primary outcomes according to pre‐published protocols in ClinicalTrials.gov and were therefore considered to be at low risk of reporting bias (AMBITION; ARIES‐1; ARIES‐2; COMPASS‐2; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN). We did not have access to protocols for the other studies.

Other potential sources of bias

It should be noted that most (16 out of 17) included studies were sponsored by the drug manufacturer.

Effects of interventions

See: Table 1

Only outcomes specified in the protocol were included in the meta‐analysis. Unless otherwise stated, data used in the review were from published material or documents released from the FDA and EMA.

Comparison 1: Endothelin receptor antagonists versus placebo

Sixteen studies reported ERA versus placebo; these have been subgrouped into selective and non‐selective ERAs.

Primary outcomes

Exercise capacity

Sixteen trials assessed change from baseline in 6MWD (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; COMPASS‐2; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1; STRIDE‐2; STRIDE‐4). The pooled data showed that participants treated with ERAs could walk 25 metres further than those treated with placebo (mean difference (MD) 25.06 m, 95% confidence interval (CI) 17.13 to 32.99; participants = 2739; studies = 14; I² = 34%; Analysis 1.1, moderate‐certainty evidence). We observed significant heterogeneity between the trials.

1.1. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 1: Change from baseline in 6‐minute walk

Subgroup analysis showed that the therapeutic effect of the non‐selective ERA (bosentan or macitentan) on 6MWD was 20.51 metres (95% CI 10.03 to 31.00; participants = 1860; studies = 8; I² = 42%), and the selective ERAs was 33.48 metres (95% CI 23.12 to 43.83; participants = 879; studies = 6; I² = 0%) compared to placebo. The test for subgroup differences was not significant (Chi² = 2.97, df = 1 (P = 0.08), I² = 66.4%).

Data from BREATHE‐2 and AMBITION were insufficient to evaluate the exercise capacity test. The BREATHE‐2 study showed a remarkable increase in the 6MWD in both treatment groups (68 metres (median) in the bosentan/prostacyclin group versus 74 metres (median) in the placebo/prostacyclin group). However, no statistical difference was observed between the two groups in the 6MWD.

AMBITION showed the median change from baseline to week 24 in 6MWD to be 48.98 (interquartile range (IQR) 4.63 to 85.75) in the combined ambrisentan plus tadalafil group and 22.70 (IQR −8.25 to 66.00) in the tadalafil‐monotherapy group, favouring the combination therapy group (P< 0.01).

WHO or NYHA functional class

Fifteen trials including a total of 3184 participants reported detailed numbers of participants with improved or deteriorated WHO/NYHA functional class (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; COMPASS‐2; EARLY; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1; STRIDE‐2; STRIDE‐4).

ERAs improved more participants' functional class than placebo (odds ratio (OR) 1.41, 95% CI 1.16 to 1.70; participants = 3060; studies = 15; I² = 5%; Analysis 1.2, moderate‐certainty evidence) and lowered the odds of functional class deterioration (OR 0.43, 95% CI 0.26 to 0.72; participants = 2347; studies = 13; I² = 40%; Analysis 1.3, moderate‐certainty evidence).

1.2. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 2: Proportion of participants with improved functional class

1.3. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 3: Proportion of participants with deteriorated functional class

For non‐selective ERAs, 248 of 1118 participants in the ERA group improved their functional class versus 117 to 778 participants in the placebo group (OR 1.45, 95% CI 1.13 to 1.87; participants = 1896; studies = 9; I² = 29%; Analysis 1.2), favouring non‐selective ERAs. For selective ERAs, 260 of 880 participants in the ERA group improved their functional class versus 90 of 408 in the placebo group (OR 1.35, 95% CI 1.01 to 1.80; participants = 1164; studies = 6; I² = 0%), favouring selective ERAs (Analysis 1.2). The test for subgroup differences was not significant (Chi² = 0.13, df = 1 (P = 0.72), I² = 0%).

Both non‐selective ERA and selective ERAs significantly lowered the occurrence of WHO/NYHA function deterioration in participants with PAH: non‐selective ERAs (OR 0.65, 95% CI 0.30 to 1.42; participants = 1121; studies = 7; I² = 32%) and selective ERAs (OR 0.31, 95% CI 0.17 to 0.60; participants = 1226; studies = 6; I² = 36%; Analysis 1.3) compared to placebo. The test for subgroup differences was not significant (Chi² = 1.98, df = 1 (P = 0.16), I² = 49.6%).

Borg dyspnoea score

Considering symptoms such as dyspnoea is an important aspect of assessing response to therapy. In the most commonly used Borg dyspnoea scale in PAH, lower is better, and the minimal clinically important difference in PAH is 1 unit (Khair 2016). Data from seven trials included in this review addressed this outcome (ARIES‐1; ARIES‐2; BREATHE‐1; Channick 2001; EDITA; SERAPH; STRIDE‐2). We found heterogeneity in the selective ERA trials that could not be explained in terms of clinical characteristics and methodological quality, so we pooled the data using a random‐effects model. Overall, ERAs may mildly reduce Borg dyspnoea scores by 0.43 units (MD −0.43, 95% CI −0.90 to 0.04; participants = 788; studies = 7; I² = 49%; Analysis 1.4, low‐certainty evidence).

1.4. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 4: Change from baseline in Borg dyspnoea index

Compared with placebo, both the non‐selective ERA bosentan (MD −0.27, 95% CI −1.58 to 1.03; participants = 240; studies = 3; I² = 60%) and selective ERAs (MD −0.43, 95% CI −1.01 to 0.14; participants = 548; studies = 4; I² = 55%; Analysis 1.4) may reduce Borg dyspnoea score. The test for subgroup differences was not significant (Chi² = 0.05, df = 1 (P = 0.83), I² = 0%).

MAESTRO reported that no relevant trends in change from baseline to week 16 in the Borg dyspnoea index existed between the macitentan and placebo groups, but did not provide detailed data. BREATHE‐5 reported Borg score as an outcome with limited detail.

AMBITION reported the treating effects in the study portal (ClinicalTrials.gov: NCT01178073) as median IQR: the change from baseline in Borg dyspnoea score was −1.00 (−2.00 to 0.50) in the combination therapy group; and −0.50 (−2.00 to 0.88) in the tadalafil‐monotherapy group.

Mortality

Twelve trials including 2889 participants reported a total of 193 deaths (AMBITION; ARIES‐1; ARIES‐2; BREATHE‐1; BREATHE‐2; COMPASS‐2; EARLY; MAESTRO; SERAPHIN; STRIDE‐1; STRIDE‐2), of which 113 were on active treatment and 80 on placebo. There was a non‐significant trend towards ERAs lowering mortality in participants with PAH (OR 0.78, 95% CI 0.58 to 1.07; participants = 2889; studies = 12; I² = 0%; Analysis 1.5, low‐certainty evidence).

1.5. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 5: Mortality

Subgroup analysis from the pooled data showed that there was no significant difference in mortality between non‐selective ERAs and selective ERAs and placebo (OR 0.88, 95% CI 0.62 to 1.23; participants = 1759; studies = 7; I² = 0%), whereas selective ERAs were found to significantly lower mortality (OR 0.45, 95% CI 0.21 to 0.94; participants = 1130; studies = 5; I² = 0%). The test for subgroup differences was not significant (Chi² = 2.62, df = 1 (P = 0.11), I² = 61.8%).

To compare the treatment effects of subgroups (selective ERAs versus non‐selective ERAs) on mortality, we calculated the interaction between the two subgroups using a method previously described by Altman 2003. The estimated interaction effect is a relative odds ratio 0.51 (95% CI 0.22 to 1.16), indicating that there is no clear evidence of a different treatment effect between non‐selective ERAs and selective ERAs on mortality.

Secondary outcomes

Cardiopulmonary haemodynamics

Ten studies measured haemodynamics variables (BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1): mean pulmonary artery pressure (mPAP), pulmonary vascular resistance (mPVR), and cardiac index. Haemodynamic variables were measured by right‐heart catheterisation in seven studies (BREATHE‐2; Channick 2001; EARLY; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1), compared with BREATHE‐1, which used an echocardiological method. Data from BREATHE‐1 were reported as a subgroup analysis in Galiè 2003.

Mean pulmonary artery pressure

Nine studies assessed mPAP (BREATHE‐2; BREATHE‐5; Channick 2001; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1). We observed significant heterogeneity between the studies, therefore we pooled them using random‐effects modelling. Pooled data from eight studies showed that ERAs lowered mPAP by 4.65 mmHg (95% CI −6.05 to −3.26; participants = 729; I² = 19%; Analysis 1.6, moderate‐certainty evidence) more than placebo (BREATHE‐5; Channick 2001; EARLY; EDITA; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1).

1.6. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 6: Change from baseline in mean pulmonary artery pressure

The non‐selective ERA (bosentan) lowered mPAP by 5.79 mmHg (95% CI −7.30 to −4.27; participants = 519; studies = 6; I² = 0%; Analysis 1.6) more than placebo, whereas the selective ERA lowered mPAP by 2.65 mmHg (95% CI −5.31 to 0.00; participants = 210; studies = 2; I² = 36%; Analysis 1.6) more than placebo. There was a significant difference in the test for subgroup differences (Chi²= 4.04, df = 1 (P = 0.04), I² = 75.2%).

BREATHE‐2 did not report data for change from baseline in mPAP, but instead as an end‐of‐treatment score: mean ± standard error (SE) was 59.2 ± 3.2 mmHg in the placebo group and 52.5 ± 2.4 mmHg in the bosentan group, which was not statistically significant (P = 0.3).

Pulmonary vascular resistance

Seven studies on 586 participants reported data of change from baseline in PVR, all of them showing that ERA significantly improved PVR compared with placebo (BREATHE‐5; Channick 2001; EARLY; EDITA; MAESTRO; PORTICO; STRIDE‐1).

We observed significant heterogeneity between studies and therefore pooled them using random‐effects modelling. Data from seven trials showed that ERAs could significantly reduce PVR by 236.24 dyn/s/cm⁵ (95% CI −333.21 to −139.26; participants = 586; studies = 7; I² = 14.7%; Analysis 1.7, moderate‐certainty evidence) more than placebo (BREATHE‐5; Channick 2001; EARLY; EDITA; MAESTRO; PORTICO; STRIDE‐1).

1.7. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 7: Change from baseline in pulmonary vascular resistance

BREATHE‐2 and SERAPHIN reported an end‐of‐treatment score, as mean and standard deviation. The pooled data showed that participants treated with ERAs had much lower PVR than those treated with placebo (MD −288.59 dyn/s/cm⁵, 95% CI −472.18 to −104.99; participants = 175; studies = 2; I² = 0%; Analysis 1.8).

1.8. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 8: Pulmonary vascular resistance

Three studies reported ratio of geometric mean PVR. The pooled data showed that participants receiving ERAs had a much lower ratio of geometric mean PVR (0.69, 95% CI 0.60 to 0.80; participants = 124; Analysis 1.9), favouring ERAs (MAESTRO; PORTICO; SERAPHIN).

1.9. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 9: Ratio of geometric mean PVR

Cardiac index

Seven studies on 718 participants reported cardiac index data (Channick 2001; EARLY; EDITA; Galiè 2003; PORTICO; SERAPHIN; STRIDE‐1). We observed significant heterogeneity between studies and therefore pooled them using a random‐effects model. Pooled data from these studies showed that ERAs could significantly increase cardiac index compared with placebo (MD 0.50 L/min/m², 95% CI 0.35 to 0.65; participants = 718; studies = 7; I² = 59%; Analysis 1.10) (Channick 2001; EARLY; EDITA; Galiè 2003; PORTICO; SERAPHIN; STRIDE‐1). The placebo‐corrected therapeutic effect of the non‐selective ERA on cardiac index was 0.55 L/min/m² (95% CI 0.34 to 0.77; participants = 509; studies = 5; I² = 70%), whilst for the selective ERA (sitaxsentan) it was 0.39 L/min/m² (95% CI 0.23 to 0.54; participants = 209; studies = 2; I² = 0%). The test for subgroup differences was not significant (Chi² = 1.50, df = 1 (P = 0.22), I² = 33.2%).

1.10. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 10: Change from baseline in cardiac index

Adverse events

Eleven trials including 2250 participants reported hepatic toxicity in detail (AMBITION; BREATHE‐1; BREATHE‐2; BREATHE‐5; Channick 2001; EARLY; MAESTRO; PORTICO; SERAPHIN; STRIDE‐1; STRIDE‐2). We observed heterogeneity between studies and therefore pooled them using a random‐effects model. Overall, 102 of 1357 participants receiving ERA had hepatic toxicity events compared with 33 of 893 participants receiving placebo, which was not significantly different (OR 1.88, 95% CI 0.91 to 3.90; participants = 2250; studies = 11; I² = 53%, moderate‐certainty evidence), though the direction of effect favoured placebo.

ARIES‐1 and ARIES‐2 reported reassuring combined data for hepatic toxicity. None of the 261 participants receiving ambrisentan developed serum aminotransferase concentrations greater than three times the upper limit of normal, compared with three participants (2.3%) receiving placebo displaying elevated levels.

Subgroup analysis

Pulmonary arterial hypertension associated with connective tissue disease

The subgroup analysis of non‐selective ERAs versus selective ERAs for mortality is described above. We could not perform a subgroup analysis comparing participants with idiopathic PAH versus those with PAH related to other conditions due to insufficient information. However, Denton 2006 reported the combined data from two trials, BREATHE‐1; Channick 2001, comparing bosentan to placebo for PAH secondary to connective tissue disease (PAH‐CTD). There was no significant difference between the bosentan and placebo groups.

Seibold 2005 reported the combined data of sitaxsentan from STRIDE‐1, STRIDE‐2, and STRIDE‐4 for PAH‐CTD, which included sitaxsentan at 50 mg, 100 mg, and 300 mg once daily; 110 of 512 participants had PAH‐CTD, including 63 with systemic sclerosis (SSc), 22 with overlap/mixed connective tissue disease, and 25 with systemic lupus erythematosus. The data showed a change from baseline in 6MWD of −16 ± 15.0 metres (mean ± SE) for placebo (N = 28); −2 ± 13.4 metres for sitaxsentan 50 mg daily (N = 26); 21 ± 10.4 metres for sitaxsentan 100 mg daily (N = 39); and 2 ± 14.1 metres for sitaxsentan 300 mg daily (N = 17). Only participants treated with 100 mg daily (the dose authorised by the EMA) significantly improved in 6MWD, which was comparable to participants with idiopathic PAH treated with sitaxsentan.

Badesch 2007 reported the pooled data for ambrisentan therapy in idiopathic PAH (n = 251) and PAH‐CTD (n = 124). At week 12, the placebo‐adjusted increase in 6MWD was 58 metres (95% CI 36 to 79) in participants with idiopathic PAH and 19 metres (95% CI −10 to 48) in participants with PAH‐CTD; the placebo‐adjusted decrease in Borg dyspnoea score was −0.8 (95% CI −1.4 to −0.19) in the idiopathic PAH group and −1.0 (95% CI −1.7 to −0.18) in the PAH‐CTD group. WHO functional class deterioration was 2.4% in participants with idiopathic PAH and 3.7% in participants with PAH‐CTD in the ambrisentan group, compared with 16.5% in participants with idiopathic PAH and 20.9% in participants with PAH‐CTD in the placebo group.

In the AMBITION study, 187 participants had CTD associated with PAH, of whom 118 had SSc associated with PAH. Initial combination therapy reduced the risk of clinical failure when compared with pooled monotherapy in each subgroup: CTD‐PAH (hazard ratio 0.43, 95% CI 0.24 to 0.77) and SSc‐PAH (0.44, 95% CI 0.22 to 0.89).

In the EDITA study, 38 participants had SSc associated with mildly elevated mPAP between 21 and 24 mmHg at rest and/or > 30 mmHg during low‐dose exercise were randomly assigned to treatment with either ambrisentan 5 to 10 mg/day or placebo. After six months, the two groups did not differ in the primary endpoint (i.e. change in mPAP). However, the ambrisentan group showed significant improvements in the secondary endpoints, that is cardiac index and PVR, at rest or at peak exercise.

Pulmonary arterial hypertension associated with portopulmonary hypertension

PORTICO first reported the effect of the non‐selective ERA macitentan on portopulmonary hypertension. The study's primary outcome (i.e. PVR at the end of study) was improved in the macitentan group, but not exercise capacity or WHO functional class at the end of study.

Pulmonary arterial hypertension associated with congenital heart disease

Two trials investigated the efficacy of non‐selective ERAs on Eisenmenger syndrome (BREATHE‐5; MAESTRO). The pooled data showed that non‐selective ERAs had no impact on exercise capacity, WHO functional class, or mortality. However, ERAs did improve mPAP (MD −4.63 mmHg, 95% CI −8.03 to −1.23; participants = 90; studies = 2; I² = 0%; Analysis 3.5) and PVR (MD −480.07 dyn/s/cm⁵, 95% CI −753.34 to −206.79; participants = 93; studies = 2; I² = 0%; Analysis 3.6).

3.5. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 5: Change from baseline in mean pulmonary arterial pressure

3.6. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 6: Change from baseline in pulmonary vascular resistance

Comparison 2: Endothelin receptor antagonists versus phosphodiesterase‐5 inhibitors

Primary outcomes

Exercise capacity

Two trials tested the efficacy of ERA versus PDE5 inhibitors (AMBITION; SERAPH).

SERAPH reported that mean 6MWD increased with both treatments compared to baseline. Sildenfafil may increase exercise capacity (MD 55 m, 95% CI 0.1 to 109.9; Analysis 2.1) when compared with bosentan in participants with PAH.

2.1. Analysis.

Comparison 2: Endothelin receptor antagonists versus PDE5 inhibitor, Outcome 1: 6‐minute walk

AMBITION showed a median change from baseline to week 24 in 6MWD of 27.00 (IQR −14.00 to 63.25) in the ambrisentan‐monotherapy group and 22.70 (IQR −8.25 to 66.00) in the tadalafil‐monotherapy group, with no difference between groups.

WHO or NYHA functional class

Only AMBITION reported the effect of ambrisentan versus tadalafil on WHO functional class. There was no difference between groups in either WHO functional improvement or deterioration (Analysis 2.2; Analysis 2.3).

2.2. Analysis.

Comparison 2: Endothelin receptor antagonists versus PDE5 inhibitor, Outcome 2: Proportion of participants with improved functional class

2.3. Analysis.

Comparison 2: Endothelin receptor antagonists versus PDE5 inhibitor, Outcome 3: Proportion of participants with deteriorated functional class

Borg dyspnoea score

Only the SERAPH study reported detailed data on Borg dyspnoea score. A trend showed that sildenafil may reduce Borg dyspnoea score compared with bosentan in participants with PAH (Analysis 2.4).

2.4. Analysis.

Comparison 2: Endothelin receptor antagonists versus PDE5 inhibitor, Outcome 4: Symptoms

Mortality

A trend showed that ERAs and PDE‐5 inhibitors may reduce mortality in participants with PAH (OR 0.32, 95% CI 0.07 to 1.36; participants = 273; studies = 2; I² = 0%; Analysis 2.5).

2.5. Analysis.

Comparison 2: Endothelin receptor antagonists versus PDE5 inhibitor, Outcome 5: Mortality

Secondary outcomes

Cardiac index

Only the SERAPH study reported data on cardiac index. There was no difference between treatments for this outcome (MD 0 L/min/m², 95% CI −0.14 to 0.14; participants = 25; studies = 1; Analysis 2.6).

2.6. Analysis.

Comparison 2: Endothelin receptor antagonists versus PDE5 inhibitor, Outcome 6: Cardiac index

Sensitivity analysis

Including or excluding combination therapy did not make a difference in the effect of ERAs on the primary outcomes (i.e. exercise capacity, functional class, and mortality) (Table 3).

2. Sensitivity analysis: including versus excluding combination therapy.

Outcome	Including combination therapy	Excluding combination therapy
Change in 6MWD, mean with 95% CI	25.06 (17.13 to 32.99)	25.65 (16.80 to 34.49)
WHO/NYHA FC improvement, OR with 95% CI	1.41 (1.16 to 1.70)	1.52 (1.22 to 1.91)
Mortality, OR with 95% CI	0.78 (0.58 to 1.07)	0.82 (0.58 to 1.17)

6MWD: 6‐minute walk distance; CI: confidence interval; NYHA FC: New York Heart Association functional class; OR: odds ratio; WHO FC: World Health Organization functional class

Including or excluding PAH participants with PAH associated with congenital heart disease and portal hypertension had little impact on the effect of ERAs on primary outcomes such as exercise capacity, functional class, and mortality (Table 4).

3. Sensitivity analysis: including versus excluding PAH participants associated with SSc, or CHD, or portal hypertension.

Outcome	Including combination therapy	Excluding combination therapy
Change in 6MWD, mean with 95% CI	25.06 (17.13 to 32.99)	27.90 (20.96 to 34.83)
WHO/NYHA FC improvement, OR with 95% CI	1.41 (1.16 to 1.70)	1.46 (1.19 to 1.78)
Mortality, OR with 95% CI	0.78 (0.58 to 1.07)	0.77 (0.57 to 1.05)

6MWD: 6‐minute walk distance; CHD: congenital heart disease; CI: confidence interval; NYHA FC: New York Heart Association functional class; OR: odds ratio; PAH: pulmonary arterial hypertension; SSc: systemic sclerosis; WHO FC: World Health Organization functional class

Discussion

The pooled data show that ERAs probably increase exercise capacity, improve WHO functional class, prevent WHO functional class deterioration, reduce dyspnoea, and improve cardiopulmonary haemodynamic variables, and may improve mortality in participants with PAH. However, they may also increase the risk of liver damage, although this was rare.

Clinical efficacy

Minimal clinically important differences (MCID) are patient‐derived scores that reflect changes in a clinical intervention that are meaningful for the patient. The six‐minute walk test (6MWD), as a reliable tool for the assessment of exercise capacity in participants with PAH, is an independent predictor of mortality (Miyamoto 2000; Sitbon 2002). Although the pooled data showed that ERAs statistically significantly increased 6MWD by 25.06 m in participants with PAH, it is less than MCID estimated by Mathai 2012. Additionally, ERAs reduced the pooled Borg dyspnoea score by 0.43, where the MCID for PAH is 1 unit (Khair 2016).

Mortality events across the included trials were low, and we did not find a significant difference for these comparisons despite the trend favouring ERAs. The included studies were underpowered for evaluating mortality, particularly given their short duration of study period and the use of time to first event of related to PAH as primary endpoint, which led to censoring of participants earlier (BREATHE‐5; Channick 2001; STRIDE‐4). There is insufficient certainty around the observed result to exclude either a reduction in mortality or no difference between groups. Transplantation or epoprostenol remains the preferred therapy with a major reduction in mortality in PAH participants with WHO functional class IV (Barst 1996).

The exact mechanism of action of ERAs on the pulmonary vascular bed remains unclear. Vasodilatation is likely to be just a part of the mechanism since usually 70% to 80% of participants with idiopathic PAH do not respond acutely to vasodilators (Rich 1992; Sitbon 1998). PAH‐targeted therapies (e.g. prostaglandins, PDE inhibitors, ERAs, and soluble guanylate cyclase stimulators), used alone or in combination, improve exercise capacity and haemodynamics and reduce hospitalisations. However, these pulmonary vasodilators do not target key features of PAH pathogenesis and have not been shown to reduce mortality (Galiè 2015; Klinger 2019; Thenappan 2018).

Using ERAs as part of combination therapy is an attractive option to target the multiple pathophysiological mechanisms in PAH. Combination therapy might be pursued by the simultaneous initiation of two (or more) treatments or by the addition of a second (or third) treatment to a previous therapy that is considered insufficient. Several combination studies of ERAs and PDE inhibitors have been carried out in participants with PAH (AMBITION; COMPASS‐2; Zhuang 2014). One randomised controlled trial used bosentan plus sildenafil (COMPASS‐2), and another two randomised controlled trials used ambrisentan plus tadalafil (AMBITION; Zhuang 2014). The evidence consistently showed that participants with PAH receiving combination therapy could walk longer than those receiving monotherapy.

The BREATHE‐2 trials, with small sample sizes, explored the efficacy and safety of combination therapy (bosentan in combination with intravenous prostacyclin), showing a trend without statistical significance towards haemodynamic and clinical improvement of this combination in participants with PAH. There is clearly an interaction between bosentan and prostacyclin in the BREATHE‐2 trial. However, in line with the predefined subgroups in the protocol for this review, the BREATHE‐2 trial was included in the meta‐analysis of ERAs versus placebo. Nevertheless, excluding BREATHE‐2 from the meta‐analysis made little difference to the results.

Subgroup analysis

Both a non‐selective ERA and a selective ERA can dilate constricted pulmonary pre‐capillary arterioles and improve exercise capacity and symptoms. Determining the relative benefit of these agents requires head‐to‐head comparisons. A post hoc analysis of the STRIDE‐2 trial that was neither designed nor powered to detect non‐inferiority or equivalence showed a mean difference in the 18‐week change from baseline between sitaxsentan and bosentan of 1.9 metres (95% CI −22.7 to 26.5; P = 0.9). Also, the AMBITION trial reported that there was no difference on 6MWD between ambrisentan‐monotherapy group and tadalafil‐monotherapy group.

ERAs showed efficacy in participants with PAH associated with CTD in post hoc analysis. ERAs did improve cardiopulmonary haemodynamic in participants with congenital heart disease and portal hypertension, but had little impact on 6MWD and functional class.

Safety and tolerability

One major concern regarding ERAs is hepatotoxicity and its capacity to cause substantial damage to the liver. This review did not show a high occurrence of hepatic toxicity. However, there were three reported cases of irreversible liver failure induced by sitaxsentan, one in the UK in 2009 and two cases from clinical trials in India and Ukraine in 2010. The reported fatal hepatic toxicity in association with sitaxsentan was idiosyncratic, unlikely to be detected by monthly monitoring, and irreversible when sitaxsentan was discontinued. Consequently, sitaxsentan was withdrawn from the market voluntarily by its licence holder. These cases emphasise the importance of dosing and hepatic monitoring in individuals with PAH.

Summary of main results

For participants with pulmonary arterial hypertension with WHO functional class II and III, endothelin receptor antagonists probably increase exercise capacity, improve WHO functional class, prevent WHO functional class deterioration, result in favourable changes in cardiopulmonary haemodynamic variables. However, they are less effective in reducing dyspnoea and mortality.

Overall completeness and applicability of evidence

All the trials identified in the current review addressed our main outcomes (i.e. 6MWD, WHO/NYHA functional class, and mortality). All relevant types of participants, interventions, and outcomes have been investigated (Table 2). The evidence we collected is highly relevant to the review questions. For people with pulmonary arterial hypertension with WHO functional class II and III, endothelin receptor antagonists probably increase exercise capacity, improve WHO functional class, prevent WHO functional class deterioration, result in favourable changes in cardiopulmonary haemodynamic variables. However, they are less effective in reducing dyspnoea and mortality.

Certainty of the evidence

All data included in the meta‐analysis were from randomised, placebo‐controlled clinical trials, and the participants, treating clinicians, and assessors were blinded. All but two trials, EARLY; EDITA, performed analyses based on intention‐to‐treat for their prespecified primary outcomes. However, the risk of attrition bias is high in the included trials due to the high occurrence of missing data (Table 2). On this basis we consider the overall certainty of the evidence to be moderate. Overall, the results of the studies included in the review are consistent, confirming the internal validity of the results of the review.

Potential biases in the review process

Many studies used Borg dyspnoea score as an outcome but provided limited data, thus bias may be introduced into the review. We analysed hepatic toxicity as an adverse event because this was listed in the protocol and approved. However, there are other serious adverse events related to ERAs which limit its use in clinical practice.

Agreements and disagreements with other studies or reviews

Our results are consistent with recent overviews (Duo‐Ji 2017; Wang 2018). Our review is more current than these overviews and provides more robust data which further confirm the clinical efficacy of ERAs in participants with PAH.

Authors' conclusions

Implications for practice.

For people with pulmonary arterial hypertension with WHO functional class II and III, endothelin receptor antagonists probably increase exercise capacity, improve WHO functional class, prevent WHO functional class deterioration, result in favourable changes in cardiopulmonary haemodynamic variables. However, they are less effective in reducing dyspnoea and mortality. Efficacy data are strongest in participants with the idiopathic form of the disease. Limited data show that combination therapy using ERAs and PDE inhibitor may provide more benefit in PAH; however, this needs to be confirmed in future studies.

Implications for research.

The question of the effects of ERAs on PAH has now likely been answered. The effect of ERAs on dyspnoea score and mortality remains uncertain, and long‐term observational studies are warranted. Combination therapy using ERAs and PDE inhibitor may provide more benefit in PAH; however, this needs to be confirmed in future studies.

What's new

Date	Event	Description
29 March 2021	Amended	Tracked changes in Figure 1 resolved.

History

Protocol first published: Issue 4, 2003
Review first published: Issue 1, 2005

Date	Event	Description
20 January 2021	New search has been performed	New literature search run, six new clinical studies added. Anticoagulation is not currently a recommended treatment for pulmonary arterial hypertension according to established guidelines, therefore we revised our inclusion criteria. We also revised our inclusion criteria with regard to the diagnosis of pulmonary arterial hypertension according to the new guidelines. We screened the trials again based on these changes.
20 January 2021	New citation required but conclusions have not changed	Adding five large‐scale clinical trials and one small trial to the review has made the evidence more robust; however, the conclusions remain unchanged. The review has been edited for style and the Plain language summary has been revised.
2 May 2013	New search has been performed	Literature search re‐run.
2 May 2013	New citation required and conclusions have changed	Substantive amendment. Five new studies added, and the results and conclusions updated.
2 May 2013	New citation required and conclusions have changed	The review has been updated with an additional two studies (Humbert 2004; Wilkins 2005). The additional data narrowed the confidence interval around the significant increase in the risk of improved functional class status in favour of endothelin receptor antagonists.

Appendices

Appendix 1. Database search strategies

CENTRAL (via the Cochrane Register of Studies)

#1 MESH DESCRIPTOR Hypertension, Pulmonary EXPLODE ALL
#2 MESH DESCRIPTOR Pulmonary Heart Disease
#3 pulmonar* near hypertensi*
#4 PAH or IPAH.ti,ab.
#5 {or #1‐#4}
#6 MESH DESCRIPTOR Receptors, Endothelin EXPLODE ALL
#7 endothelin* near receptor*
#8 endothelium* near receptor*
#9 bosentan
#10 Sitaxentan
#11 ambrisentan
#12 Macitentan
#13 Selexipag
#14 {or #6‐#13}
#15 #14 AND #5

MEDLINE Ovid SP

1. exp Hypertension, Pulmonary/
2. Pulmonary Heart Disease/
3. (pulmonar* adj2 hypertensi*).tw.
4. (PAH or IPAH).ab,ti.
5. or/1‐4
6. exp Receptors, Endothelin/
7. exp Endothelin Receptor Antagonists/
8. ((endothelin$ or endothelium$) adj2 receptor$).tw.
9. bosentan.tw.
10. Sitaxentan.tw.
11. ambrisentan.tw.
12. Macitentan.tw.
13. Selexipag.tw.
14. or/6‐13
15. 5 and 14
16. (controlled clinical trial or randomised controlled trial).pt.
17. (randomised or randomised).ab,ti.
18. placebo.ab,ti.
19. dt.fs.
20. randomly.ab,ti.
21. trial.ab,ti.
22. groups.ab,ti.
23. or/16‐22
24. Animals/
25. Humans/
26. 24 not (24 and 25)
27. 23 not 26
28. 15 and 27

EMBASE Ovid SP

1. exp Pulmonary Hypertension/
2. (pulmonar* adj2 hypertensi*).tw.
3. (PAH or IPAH).ab,ti.
4. or/1‐3
5. exp endothelin receptor antagonist/
6. ((endothelin$ or endothelium$) adj2 receptor$).tw.
7. bosentan.tw.
8. Sitaxentan.tw.
9. ambrisentan.tw.
10. Macitentan.tw.
11. Selexipag.tw.
12. or/5‐11
13. 4 and 12
14. Randomized Controlled Trial/
15. randomisation/
16. controlled clinical trial/
17. Double Blind Procedure/
18. Single Blind Procedure/
19. Crossover Procedure/
20. (clinica$ adj3 trial$).tw.
21. ((singl$ or doubl$ or trebl$ or tripl$) adj3 (mask$ or blind$ or method$)).tw.
22. exp Placebo/
23. placebo$.ti,ab.
24. random$.ti,ab.
25. ((control$ or prospectiv$) adj3 (trial$ or method$ or stud$)).tw.
26. (crossover$ or cross‐over$).ti,ab.
27. or/14‐26
28. exp animals/ or exp invertebrate/ or animal experiment/ or animal model/ or animal tissue/ or animal cell/ or nonhuman/
29. human/ or normal human/ or human cell/
30. 28 and 29
31. 28 not 30
32. 27 not 31
33. 13 and 32

ClinicalTrials.gov

Condition	Pulmonary Arterial Hypertension OR pulmonary hypertension
intervention	bosentan OR Sitaxentan OR ambrisentan OR Macitentan OR Selexipag
study type	Interventional

WHO ICTRP

Condition	Pulmonary Arterial Hypertension OR pulmonary hypertension
intervention	bosentan OR Sitaxentan OR ambrisentan OR Macitentan OR Selexipag

Data and analyses

Comparison 1. Endothelin receptor antagonists versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Change from baseline in 6‐minute walk	14	2739	Mean Difference (IV, Random, 95% CI)	25.06 [17.13, 32.99]
1.1.1 Non‐selective ERA	8	1860	Mean Difference (IV, Random, 95% CI)	20.51 [10.03, 31.00]
1.1.2 Selective ERA	6	879	Mean Difference (IV, Random, 95% CI)	33.48 [23.12, 43.83]
1.2 Proportion of participants with improved functional class	15	3060	Odds Ratio (M‐H, Fixed, 95% CI)	1.41 [1.16, 1.70]
1.2.1 Non‐selective ERAs	9	1896	Odds Ratio (M‐H, Fixed, 95% CI)	1.45 [1.13, 1.87]
1.2.2 Selective ERAs	6	1164	Odds Ratio (M‐H, Fixed, 95% CI)	1.35 [1.01, 1.80]
1.3 Proportion of participants with deteriorated functional class	13	2347	Odds Ratio (M‐H, Random, 95% CI)	0.43 [0.26, 0.72]
1.3.1 Non‐selective ERA	7	1121	Odds Ratio (M‐H, Random, 95% CI)	0.65 [0.30, 1.42]
1.3.2 Selective ERAs	6	1226	Odds Ratio (M‐H, Random, 95% CI)	0.31 [0.17, 0.60]
1.4 Change from baseline in Borg dyspnoea index	7	788	Mean Difference (IV, Random, 95% CI)	‐0.43 [‐0.90, 0.04]
1.4.1 Non‐selective ERAs	3	240	Mean Difference (IV, Random, 95% CI)	‐0.27 [‐1.58, 1.03]
1.4.2 Selective ERAs	4	548	Mean Difference (IV, Random, 95% CI)	‐0.43 [‐1.01, 0.14]
1.5 Mortality	12	2889	Odds Ratio (M‐H, Fixed, 95% CI)	0.78 [0.58, 1.07]
1.5.1 Non‐selective ERAs	7	1759	Odds Ratio (M‐H, Fixed, 95% CI)	0.88 [0.62, 1.23]
1.5.2 Selective ERAs	5	1130	Odds Ratio (M‐H, Fixed, 95% CI)	0.45 [0.21, 0.94]
1.6 Change from baseline in mean pulmonary artery pressure	8	729	Mean Difference (IV, Random, 95% CI)	‐4.65 [‐6.05, ‐3.26]
1.6.1 Non‐selective ERAs	6	519	Mean Difference (IV, Random, 95% CI)	‐5.79 [‐7.30, ‐4.27]
1.6.2 Selective ERAs	2	210	Mean Difference (IV, Random, 95% CI)	‐2.65 [‐5.31, 0.00]
1.7 Change from baseline in pulmonary vascular resistance	7	586	Mean Difference (IV, Random, 95% CI)	‐236.24 [‐333.21, ‐139.26]
1.7.1 Non‐selective ERAs	5	376	Mean Difference (IV, Random, 95% CI)	‐281.74 [‐395.85, ‐167.63]
1.7.2 Selective ERAs	2	210	Mean Difference (IV, Random, 95% CI)	‐173.73 [‐332.52, ‐14.94]
1.8 Pulmonary vascular resistance	2	175	Mean Difference (IV, Fixed, 95% CI)	‐288.59 [‐472.18, ‐104.99]
1.9 Ratio of geometric mean PVR	3		Ratio of Geometric mean (IV, Random, 95% CI)	Subtotals only
1.9.1 Selective ERAs	0		Ratio of Geometric mean (IV, Random, 95% CI)	Not estimable
1.9.2 Non‐selective ERAs	3		Ratio of Geometric mean (IV, Random, 95% CI)	0.69 [0.60, 0.80]
1.10 Change from baseline in cardiac index	7	718	Mean Difference (IV, Random, 95% CI)	0.50 [0.35, 0.65]
1.10.1 Non‐selective ERAs	5	509	Mean Difference (IV, Random, 95% CI)	0.55 [0.34, 0.77]
1.10.2 Selective ERAs	2	209	Mean Difference (IV, Random, 95% CI)	0.39 [0.23, 0.54]
1.11 Change from baseline in SpO 2	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
1.12 Hepatic toxicity	11	2250	Odds Ratio (M‐H, Random, 95% CI)	1.88 [0.91, 3.90]
1.12.1 Non‐selective ERAs	9	1888	Odds Ratio (M‐H, Random, 95% CI)	2.33 [0.98, 5.56]
1.12.2 Selective ERAs	2	362	Odds Ratio (M‐H, Random, 95% CI)	0.89 [0.31, 2.51]

1.11. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 11: Change from baseline in SpO ₂

1.12. Analysis.

Comparison 1: Endothelin receptor antagonists versus placebo, Outcome 12: Hepatic toxicity

Comparison 2. Endothelin receptor antagonists versus PDE5 inhibitor.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 6‐minute walk	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.2 Proportion of participants with improved functional class	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.3 Proportion of participants with deteriorated functional class	1		Odds Ratio (M‐H, Random, 95% CI)	Totals not selected
2.4 Symptoms	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.5 Mortality	2	273	Odds Ratio (M‐H, Fixed, 95% CI)	0.32 [0.07, 1.36]
2.6 Cardiac index	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 3. Endothelin receptor antagonists in Eisenmenger syndrome.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Change from baseline in 6‐minute walk	2	280	Mean Difference (IV, Random, 95% CI)	21.49 [‐31.23, 74.21]
3.1.1 Non‐selective ERA	2	280	Mean Difference (IV, Random, 95% CI)	21.49 [‐31.23, 74.21]
3.1.2 Selective ERA	0	0	Mean Difference (IV, Random, 95% CI)	Not estimable
3.2 Proportion of participants with improved functional class	2	280	Odds Ratio (M‐H, Fixed, 95% CI)	0.95 [0.48, 1.90]
3.2.1 Non‐selective ERAs	2	280	Odds Ratio (M‐H, Fixed, 95% CI)	0.95 [0.48, 1.90]
3.2.2 Selective ERAs	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	Not estimable
3.3 Proportion of participants with deteriorated functional class	2	280	Odds Ratio (M‐H, Random, 95% CI)	0.67 [0.09, 4.85]
3.3.1 Non‐selective ERA	2	280	Odds Ratio (M‐H, Random, 95% CI)	0.67 [0.09, 4.85]
3.3.2 Selective ERAs	0	0	Odds Ratio (M‐H, Random, 95% CI)	Not estimable
3.4 Mortality	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.4.1 Non‐selective ERAs	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.4.2 Selective ERAs	0		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.5 Change from baseline in mean pulmonary arterial pressure	2	90	Mean Difference (IV, Fixed, 95% CI)	‐4.63 [‐8.03, ‐1.23]
3.5.1 Non‐selective ERAs	2	90	Mean Difference (IV, Fixed, 95% CI)	‐4.63 [‐8.03, ‐1.23]
3.5.2 Selective ERAs	0	0	Mean Difference (IV, Fixed, 95% CI)	Not estimable
3.6 Change from baseline in pulmonary vascular resistance	2	93	Mean Difference (IV, Fixed, 95% CI)	‐480.07 [‐753.34, ‐206.79]
3.6.1 Non‐selective ERAs	2	93	Mean Difference (IV, Fixed, 95% CI)	‐480.07 [‐753.34, ‐206.79]
3.6.2 Selective ERAs	0	0	Mean Difference (IV, Fixed, 95% CI)	Not estimable
3.7 Change from baseline in SpO 2	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

3.1. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 1: Change from baseline in 6‐minute walk

3.2. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 2: Proportion of participants with improved functional class

3.3. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 3: Proportion of participants with deteriorated functional class

3.4. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 4: Mortality

3.7. Analysis.

Comparison 3: Endothelin receptor antagonists in Eisenmenger syndrome, Outcome 7: Change from baseline in SpO ₂

Characteristics of studies

Characteristics of included studies [ordered by study ID]

AMBITION.

Study characteristics
Methods	Study design: Randomised controlled trial Study grouping: Parallel group
Participants	Baseline characteristics Combination (ambrisentan plus tadalafil) therapy No. randomised to group: 253 Age (mean ± SD): 54.5 ± 14.3 Female sex (%): 188 (74) Hypertension (%): 104 (41) Diabetes (%): 19 (8) CAD (%): 16 (6) Idiopathic PAH (%): 127 (50) Heritable PAH (%): 7 (3) PAH associated with CTD (%): 103 (41) PAH associated with CHD (%): 5 (2) PAH associated with human immunodeficiency (%): 5 (2) PAH associated with drug use or toxin exposure (%): 6 (2) PAP mmHg (mean ± SD): 48.1 ± 12.4 Cardiac index, L/min/m2 (mean ± SD): 2.41 ± 0.64 PVR, dyn/s/cm5 (mean ± SD): 824.1 ± 467.0 6MWD, m (mean ± SD): 353.5 ± 87.9 WHO FC II (%): 76 (30) WHO FC III (%): 177 (70) Ambrisentan monotherapy No. randomised to group: 126 Age (mean ± SD): 53.9 ± 14.7 Female sex (%): 100 (79) Hypertension (%): 52 (41) Diabetes (%): 13 (10) CAD (%): 2 (2) Idiopathic PAH (%): 72 (57) Heritable PAH (%): 3 (2) PAH associated with CTD (%): 44 (35) PAH associated with CHD (%): 1 (1) PAH associated with human immunodeficiency (%): 2 (2) PAH associated with drug use or toxin exposure (%): 4 (3) PAP, mmHg (mean ± SD): 50.4 ± 12.5 Cardiac index, L/min/m2 (mean ± SD): 2.41 ± 0.66 PVR, dyn/s/cm5 (mean ± SD): 852.4 ± 394.7 6MWD, m (mean ± SD): 354.2 ± 92.3 WHO FC II (%): 38 (30) WHO FC III (%): 88 (70) Tadalafil monotherapy No. randomised to group: 121 Age (mean ± SD): 54.5 ± 15.2 Female sex (%): 100 (83) Hypertension (%): 43 (36) Diabetes (%): 17 (14) CAD (%): 2 (2) Idiopathic PAH (%): 66 (55) Heritable PAH (%): 4 (3) PAH associated with CTD (%): 40 (33) PAH associated with CHD (%): 3 (2) PAH associated with human immunodeficiency (%): 2 (2) PAH associated with drug use or toxin exposure (%): 6 (5) PAP, mmHg (mean ± SD): 48.1 ± 12.6 Cardiac index, L/min/m2 (mean ± SD): 2.45 ± 0.77 PVR, dyn/s/cm5 (mean ± SD): 798.0 ± 409.4 6MWD, m (mean ± SD): 349.2 ± 91.6 WHO FC II (%): 41 (34) WHO FC III (%): 80 (66)
Interventions	Intervention characteristics Combination (ambrisentan plus tadalafil) therapy Timing: once daily Co‐interventions: anticoagulant, calcium‐channel blocker, diuretic, aldosterone antagonist Compliance: high Duration of treatment period: 24 weeks Ambrisentan monotherapy Timing: once daily Co‐interventions: anticoagulant, calcium‐channel blocker, diuretic, aldosterone antagonist Compliance: high Duration of treatment period: 24 weeks Tadalafil monotherapy Timing: once daily Co‐interventions: anticoagulant, calcium‐channel blocker, diuretic, aldosterone antagonist Compliance: high Duration of treatment period: 24 weeks
Outcomes	6MWD Mortality WHO FC improved WHO FC deteriorated
Identification	Sponsorship source: Gilead Sciences and GlaxoSmithKline Country: Italy Setting: International, multicentre study Comments: p. 843, the end of study Author's name: Nazzareno Galiè Institution: Department of Experimental, Diagnostic, and Specialty Medicine, University of Bologna Email: nazzareno.galie@unibo.it Address: Massarenti 9, Bologna 40138, Italy
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Judgement comment: Randomisation was performed centrally by the study sponsors using an interactive voice‐response system. p. 836, the first paragraph in section "study procedures"
Allocation concealment (selection bias)	Low risk	"Matching placebo tablets were administered to maintain blinding" in appendix materials. p. 22
Blinding (performance bias and detection bias) All outcomes	Low risk	Judgement comment: "Matching placebo tablets were administered to maintain blinding" in appendix materials. p. 22
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing in 3 intervention groups and missing data were not balanced. 21.7% (55/253) participants discontinued from the combination therapy group, 34.1% (43/126) participants from the ambrisentan group, and 24.0% (29/121) participants from the tadalafil group.
Selective reporting (reporting bias)	Low risk	Reported as per protocol published at ClinicalTrials.gov.

ARIES‐1.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted centrally according to a computer‐generated random number.
Participants	Baseline characteristics Placebo n = 67 Age (mean ± SD): 48 ± 16 Female sex (%): 59 (88) Idiopathic PAH (%): 43 (64) PAH associated with CTD (%): 21 (31) PAH associated with human immunodeficiency (%): 2 (3) PAH associated with drug use or toxin exposure (%): 1 (2) PAP mmHg (mean ± SD): 50 ± 15 Cardiac index, L/min/m2 (mean ± SD): 2.5 ± 0.8 PVR, dyn/s/cm5 (mean ± SD): 868 ± 518 6MWD, m (mean ± SD): 342 ± 73 WHO FC I (%): 2 (3) WHO FC II (%): 23 (34) WHO FC III (%): 41 (46) WHO FC IV (%): 1 (2) Ambrisentan (5 mg/day) n = 67 age (mean ± SD): 53 ± 14 Female sex (%): 56 (84) Idiopathic PAH (%): 42 (63) PAH associated with CTD (%): 19 (28) PAH associated with human immunodeficiency (%): 3 (5) PAH associated with drug use or toxin exposure (%): 2 (3) PAP mmHg (mean ± SD): 47 ± 13 Cardiac index, L/min/m2 (mean ± SD): 2.5 ± 0.9 PVR, dyn/s/cm5 (mean ± SD): 834 ± 424 6MWD, m (mean ± SD): 340 ± 77 WHO FC I (%): 1 (2) WHO FC II (%): 20 (30) WHO FC III (%): 40 (60) WHO FC IV (%): 6 (9) Ambrisentan (10 mg/day) n = 67 Age (mean ± SD): 49 ± 16 Female sex (%): 53 (79) Idiopathic PAH (%): 41 (61) PAH associated with CTD (%): 22 (33) PAH associated with human immunodeficiency (%): 2 (3) PAH associated with drug use or toxin exposure (%): 2 (3) PAP mmHg (mean ± SD): 51 ± 16 Cardiac index, L/min/m2 (mean ± SD): 2.6 ± 0.7 PVR, dyn/s/cm5 (mean ± SD): 912 ± 465 6MWD, m (mean ± SD): 341 ± 78 WHO FC I (%): 2 (3) WHO FC II (%): 22 (33) WHO FC III (%): 36 (54) WHO FC IV (%): 7 (10)
Interventions	Participants were randomised to receive placebo or ambrisentan 5 or 10 mg orally daily for 12 weeks. Timing: once daily Co‐interventions: bosentan, sitaxsentan, sildenafil, epoprostenol, iloprost, or treprostinil was prohibited. Compliance: high Duration of treatment period: 12 weeks
Outcomes	Primary outcomes: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening, plasma BNP and 36‐item Short Form Health Survey (SF‐36) physical functioning scale
Identification	‐
Notes	The combined data of ARIES‐1 and ARIES‐2 showed that 21 participants discontinued in the placebo group in ARIES‐1 and ARIES‐2, 6 participants in the ambrisentan 2.5 mg group, 9 participants in the ambrisentan 5 mg group, and 5 participants in the ambrisentan 10 mg group.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"A central randomisation scheme stratified by PAH cause (idiopathic versus other PAH causes) was used to assign participants."
Allocation concealment (selection bias)	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing in the intervention groups, and missing data were not balanced. A total of 41 participants in ARIES‐1 and ARIES‐2 studies discontinued prematurely during the 12‐week treatment period: 21 (15.9%) receiving placebo and 20 (7.6%) receiving ambrisentan.
Selective reporting (reporting bias)	Low risk	Prespecified primary outcome was reported in the results. ClinicalTrials.gov Identifier: NCT00091598

ARIES‐2.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted centrally according to a computer‐generated random number.
Participants	Baseline characteristics Placebo n = 65 Age (mean ± SD): 51 ± 14 Female sex (%): 44 (68) Idiopathic PAH (%): 42 (65) PAH associated with CTD (%): 22 (34) PAH associated with human immunodeficiency (%): 1 (2) PAH associated with drug use or toxin exposure (%): 0 (0) PAP mmHg (mean ± SD): 51 ± 13 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.7 PVR, dyn/s/cm5 (mean ± SD): 971 ± 579 6MWD, m (mean ± SD): 343 ± 86 WHO FC I (%): 2 (3) WHO FC II (%): 24 (37) WHO FC III (%): 37 (57) WHO FC IV (%): 2 (3) Ambrisentan (5 mg/day) n = 64 Age (mean ± SD): 52 ± 15 Female sex (%): 48 (75) Idiopathic PAH (%): 42 (66) PAH associated with CTD (%): 19 (30) PAH associated with human immunodeficiency (%): 2 (3) PAH associated with drug use or toxin exposure (%): 1 (2) PAP mmHg (mean ± SD): 48 ± 14 Cardiac index, L/min/m2 (mean ± SD): 2.5 ± 0.7 PVR, dyn/s/cm5 (mean ± SD): 800 ± 396 6MWD, m (mean ± SD): 347 ± 84 WHO FC I (%): 0 (0) WHO FC II (%): 34 (53) WHO FC III (%): 29 (45) WHO FC IV (%): 1 (2) Ambrisentan (5 mg/day) n = 63 Age (mean ± SD): 50 ± 16 Female sex (%): 51 (81) Idiopathic PAH (%): 41 (65) PAH associated with CTD (%): 21 (33) PAH associated with human immunodeficiency (%): 1 (2) PAH associated with drug use or toxin exposure (%): 0 (0) PAP mmHg (mean ± SD): 48 ± 14 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.8 PVR, dyn/s/cm5 (mean ± SD): 931 ± 672 6MWD, m (mean ± SD): 355 ± 84 WHO FC I (%): 1 (2) WHO FC II (%): 28 (44) WHO FC III (%): 33 (52) WHO FC IV (%): 1 (2)
Interventions	Participants were randomised to receive placebo or ambrisentan 2.5 or 5 mg daily for 12 weeks. Timing: once daily Co‐interventions: bosentan, sitaxsentan, sildenafil, epoprostenol, iloprost, or treprostinil was prohibited. Compliance: high Duration of treatment period: 12 weeks
Outcomes	Primary outcomes: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening, plasma BNP and 36‐item Short Form Health Survey (SF‐36) physical functioning scale
Identification	‐
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"A central randomisation scheme stratified by PAH cause (idiopathic versus other PAH causes) was used to assign participants."
Allocation concealment (selection bias)	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing in the intervention groups, and missing data were not balanced. A total of 41 participants in ARIES‐1 and ARIES‐2 studies discontinued prematurely during the 12‐week treatment period: 21 (15.9%) receiving placebo and 20 (7.6%) receiving ambrisentan.
Selective reporting (reporting bias)	Low risk	Prespecified primary outcome was reported in the results. ClinicalTrials.gov Identifier: NCT00091598

BREATHE‐1.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted according to a computer‐generated random number.
Participants	Baseline characteristics Placebo n = 69 Age (mean ± SD): 47.2 ± 16.2 Female sex (%): 54 (78) Idiopathic PAH (%): 48 (70) PAH associated with CTD (%): 21 (30) PAP mmHg (mean ± SD): 53 ± 17 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.7 PVR, dyn/s/cm5 (mean ± SD): 880 ± 540 6MWD, m (mean ± SD): 344 ± 76 WHO FC III (%): 65 (94) WHO FC IV (%): 4 (6) Bosentan n = 144 Age (mean ± SD): 48.7 ± 15.8 Female sex (%): 114 (79) Idiopathic PAH (%): 102 (71) PAH associated with CTD (%): 32 (29) PAP mmHg (mean ± SD): 55 ± 16 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.8 PVR, dyn/s/cm5 (mean ± SD): 1014 ± 678 6MWD, m (mean ± SD): 330 ± 74 WHO FC III (%): 130 (90) WHO FC IV (%): 14 (10)
Interventions	Intervention group: 62.5 mg bosentan twice daily for the first 4 weeks followed by the target dose (125 mg or 250 mg twice daily) for 12 weeks Control group: placebo Timing: twice daily Co‐interventions: antithrombotic agents, diuretics, and calcium‐channel blockers Compliance: high Duration of treatment period: 16 weeks
Outcomes	Primary outcomes: change from baseline in 6MWD Secondary outcomes: Borg dyspnoea index; WHO FC
Identification
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised allocation by pharmacy‐controlled randomisation
Allocation concealment (selection bias)	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	9% participants discontinued from the placebo group due to suffering clinical worsening of symptoms of PAH or syncope occurred, as compared with no participants in the bosentan group. The missing data were not balanced between groups.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

BREATHE‐2.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted according to a computer‐generated random number.
Participants	Baseline characteristics Placebo/epoprostenol n = 11 Age (mean ± SD): 47 ± 19 Female sex (%): 6 (55) Idiopathic PAH (%): 10 (91) PAH associated with CTD (%): 1 (9) PAP mmHg (mean ± SD): 60.9 ± 2.9 Cardiac index, L/min/m2 (mean ± SD): 1.7 ± 0.2 PVR, dyn/s/cm5 (mean ± SD): 1426 ± 140 WHO FC III (%): 8 (73) WHO FC IV (%): 3 (17) Bosentan/epoprostenol n = 22 Age (mean ± SD): 45 ± 17 Female sex (%): 17 (77) Idiopathic PAH (%): 17 (77) PAH associated with CTD (%): 4 (23) PAP mmHg (mean ± SD): 59.2 ± 4.0 Cardiac index, L/min/m2 (mean ± SD): 1.7 ± 0.1 PVR, dyn/s/cm5 (mean ± SD): 1511 ± 129 WHO FC III (%): 17 (77) WHO FC IV (%): 4 (23)
Interventions	33 participants with PAH started prostacyclin treatment (2 ng/kg/min starting dose, up to 14 ± 2 ng/kg/min at week 16) and were randomised for 16 weeks in a 2:1 ratio to bosentan (62.5 mg twice daily for 4 weeks then 125 mg twice daily) or placebo. Timing: once daily Co‐interventions: antithrombotic agents, diuretics Compliance: high Duration of treatment period: 16 weeks
Outcomes	Primary outcomes: change from baseline to week 16 in TPR Secondary outcomes: change in cardiac index, PVR, mPAP, and mRAP, 6MWD, NYHA FC and dyspnoea‐fatigue rating
Identification	‐
Notes	The standard deviations of change from baseline in 6MWD were estimated from figures.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised allocation by pharmacy‐controlled randomisation
Allocation concealment (selection bias)	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing (1/11 in the placebo group versus 4/22 in the bosentan group), and missing data were not balanced.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

BREATHE‐5.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted according to a computer‐generated random number.
Participants	Baseline characteristics Placebo n = 17 age (mean ± SD): 44.2 ± 8.5 Female sex (%): 10 (59) PAH associated with CHD (%): 17 (100) PAP mmHg (mean ± SD): 72.1 ± 19.4 PVR, dyn/s/cm5 (mean ± SD): 2870 ± 1209 6MWD, m (mean ± SD): 366.4 ± 67.5 WHO FC III (%): 17 (100) Bosentan n = 37 age (mean ± SD): 37.2 ± 12.0 Female sex (%): 23 (62) PAH associated with CHD (%): 37 (100) PAP mmHg (mean ± SD): 77.8 ± 15.2 PVR, dyn/s/cm5 (mean ± SD): 3425 ± 1410 6MWD, m (mean ± SD): 331.9 ± 82.8 WHO FC III (%): 37 (100)
Interventions	Intervention group: 62.5 mg bosentan twice daily for the first 4 weeks followed by the target dose (125 mg twice daily) for 12 weeks. Participants who did not tolerate the target dose of 125 mg twice daily could be down titrated to the starting dose (62.5 mg twice daily). Control group: placebo Timing: once daily Co‐interventions: prostanoids, phosphodiesterase‐V inhibitors, and endothelin receptor antagonists were not allowed. Compliance: high Duration of treatment period: 16 weeks
Outcomes	Primary outcomes: change from baseline in SpO2 and PVR index Secondary outcomes: change from baseline in 6MWD, WHO FC, cardiac index, PVR, PAP, and RAP
Identification	‐
Notes	Standard deviations were estimated from standard errors.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was controlled by study medication packaging (Almedica HPS AG, Reinach, Switzerland)."
Allocation concealment (selection bias)	Low risk	"The investigators, participants, monitors, and sponsor personnel remained blinded to the treatment until closure of the clinical database." p. 49
Blinding (performance bias and detection bias) All outcomes	Low risk	"The investigators, participants, monitors, and sponsor personnel remained blinded to the treatment until closure of the clinical database." p. 49
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing. 2/17 (11.8%) in the placebo group versus 2/37 (5.4%) in the bosentan group discontinued from the study. Missing data were balanced.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

Channick 2001.

Study characteristics
Methods	Multicentre, randomised, parallel trial comparing bosentan with placebo for 12 weeks. Randomisation was computer generated with a block size of 3.
Participants	Baseline characteristics Placebo n = 11 Age (mean ± SD): 47.4 ± 14.0 Female sex (%): 11 (100) Idiopathic PAH (%): 10 (91) PAH associated with CTD (%): 1 (9) PAP mmHg (mean ± SD): 56 ± 10 Cardiac index, L/min/m2 (mean ± SD): 2.5 ± 1.0 PVR, dyn/s/cm5 (mean ± SD): 942 ± 430 6MWD, m (mean ± SD): 355 ± 82 WHO FC III (%): 11 (100) Bosentan n = 21 Age (mean ± SD): 52.2 ± 12.2 Female sex (%): 18 (81) Idiopathic PAH (%): 17 (81) PAH associated with CTD (%): 4 (19) PAP mmHg (mean ± SD): 54 ± 13 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.7 PVR, dyn/s/cm5 (mean ± SD): 896 ± 425 6MWD, m (mean ± SD): 360 ± 86 WHO FC III (%): 21 (100)
Interventions	Intervention group: 62.5 mg bosentan twice daily for the first 4 weeks followed by the target dose (125 mg twice daily) Control group: placebo Timing: twice daily Co‐interventions: warfarin and calcium channel blockers Compliance: high Duration of treatment period: 16 weeks
Outcomes	Primary outcomes: 6MWD Secondary outcomes: cardiopulmonary haemodynamics (cardiac index, PVR, mPAP, and mRAP), WHO FC
Identification
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was computer generated using Drug Labelling System with a block size of three."
Allocation concealment (selection bias)	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing (2/11) in the placebo group, but not in the bosentan group (0/21). Missing data were not balanced.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

COMPASS‐2.

Study characteristics
Methods	Study design: Randomised controlled trial Study grouping: Parallel group
Participants	Baseline characteristics Bosentan n = 159 Sex ‐ female (%): 125 (78.6) Age years, mean ± SD: 52.9 ± 15.4 Idiopathic PAH (%): 99 (62.3) Familial PAH (%): 3 (1.9) PAH secondary to CTD (%): 43 (27.0) PAH associated with CHD (%): 9 (5.7) PAH secondary to drugs and toxins (%): 5 (3.1) 6MWD, m, mean ± SD: 363.1 ± 78.5 WHO FC II (%): 71 (44.7) WHO FC III (%): 88 (55.3) WHO FC IV (%): 0 (0) Baseline sildenafil dose mg, mean ± SD: 77.8 ± 48.5 Placebo n = 175 Sex ‐ female (%): 128 (73.1) Age years, mean ± SD: 54.7 ± 15.7 Idiopathic PAH (%): 114 (65.1) Familial PAH (%): 2 (1.1) PAH secondary to CTD (%): 45 (25.7) PAH associated with CHD (%): 11 (6.3) PAH secondary to drugs and toxins (%): 3 (1.7) 6MWD, m, mean ± SD: 357.6 ± 73.1 WHO FC (%): 69 (39.4) WHO FC III (%): 104 (59.4) WHO FC IV (%): 2 (1.1) Baseline sildenafil dose mg, mean ± SD: 81.1 ± 45.1
Interventions	Intervention characteristics Bosentan Timing: twice daily Co‐interventions: sildenafil Compliance: high Duration of treatment period: 16 weeks Placebo Timing: twice daily Co‐interventions: sildenafil Compliance: high Duration of treatment period: 16 weeks
Outcomes	Mortality WHO FC improved WHO FC worsened Change in 6MWD Hepatic toxicity
Identification	Sponsorship source: Actelion Pharmaceuticals Ltd. Setting: International, multicentre, university‐based hospital Author's name: Vallerie McLaughlin Institution: University of Michigan Health System Email: vmclaugh@umich.edu Address: Department of Internal Medicine, Division of Cardiovascular Medicine, University of Michigan Health System, Ann Arbor, MI 48109‐5853, USA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"COMPASS‐2 was a prospective, international, randomised, double‐blind, placebo‐controlled, event‐driven trial"
Allocation concealment (selection bias)	Low risk	"COMPASS‐2 was a prospective, international, randomised, double‐blind, placebo‐controlled, event‐driven trial (www.clinicaltrials.gov identifier number NCT00303459)."
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	25.7% participants were discontinued from the placebo group and 31.4% from the bosentan group.
Selective reporting (reporting bias)	Low risk	Reported according to the protocol published at ClinicalTrials.gov.

EARLY.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial
Participants	Baseline characteristics Placebo n = 92 Age (mean ± SD): 44.2 ± 16.5 Female sex (%): 58 (63) Idiopathic PAH (%): 58 (63) PAH associated with CTD (%): 16 (17) PAH associated with CHD (%): 15 (16) PAH associated with human immunodeficiency (%): 2 (2) PAP mmHg (mean ± SD): 52.3 ± 16.0 Cardiac index, L/min/m2 (mean ± SD): 2.7 ± 0.6 PVR, dyn/s/cm5 (mean ± SD): 805 ± 369 6MWD, m (mean ± SD): 431 ± 91 WHO FC II (%): 92 (100) Bosentan n = 93 Age (mean ± SD): 45.2 ± 17.9 Female sex (%): 71 (76) Idiopathic PAH (%): 54 (58) PAH associated with CTD (%): 16 (17) PAH associated with CHD (%): 18 (19) PAH associated with human immunodeficiency (%): 5 (5) PAP mmHg (mean ± SD): 52.5 ± 18.9 Cardiac index, L/min/m2 (mean ± SD): 2.7 ± 0.8 PVR, dyn/s/cm5 (mean ± SD): 839 ± 531 6MWD, m (mean ± SD): 438 ± 86 WHO FC II (%): 93 (100)
Interventions	Participants treated with either bosentan at an initial dose of 62.5 mg twice daily, uptitrating to 125 mg twice daily after 4 weeks, or remaining at 62.5 mg twice daily if bodyweight < 40 kg, or placebo for a 6‐month double‐blind treatment period. Timing: twice daily Co‐interventions: oral anticoagulants, sildenafil, and calcium channel blockers Compliance: high Duration of treatment period: 6 months
Outcomes	Primary outcomes: PVR and change from baseline in 6MWD Secondary outcomes: time to clinical worsening, change from baseline to month 6 in WHO FC, Borg dyspnoea index, mPAP, cardiac index, RAP, and mixed venous oxygen saturation
Identification	‐
Notes	Standard deviations were estimated from confidence intervals.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Participants were randomly assigned in a one to one ratio to receive either bosentan using a centralised integrated voice recognition system."
Allocation concealment (selection bias)	Low risk	"This system assigned a unique randomisation number to each participant and designated the correct blinded study medication to be dispensed, both at the start of study treatment and at each scheduled visit. This code was accessible only to authorised individuals who were not involved in the conduct or analysis of the study, until the time of unblinding."
Blinding (performance bias and detection bias) All outcomes	Low risk	"This system assigned a unique randomisation number to each participant and designated the correct blinded study medication to be dispensed, both at the start of study treatment and at each scheduled visit. This code was accessible only to authorised individuals who were not involved in the conduct or analysis of the study, until the time of unblinding."
Incomplete outcome data (attrition bias) All outcomes	Low risk	12.9% (12/93) participants discontinued bosentan, and 10.8% (10/92) participants discontinued placebo. Missing data were balanced.
Selective reporting (reporting bias)	Low risk	Prespecified primary outcome reported in the results. ClinicalTrials.gov Identifier: NCT00091715

EDITA.

Study characteristics
Methods	Study design: Randomised controlled trial Study grouping: Parallel group
Participants	Baseline characteristics Placebo n = 19 Female sex (%): 14 (73.7) Age ‐ year, mean ± SD: 54.89 ± 11.23 PAH associated with CTD (%): 19 (100.0) 6MWD, m, mean ± SD: 448.11 ±82.64 WHO FC II (%): 15 (78.9) WHO FC III (%): 4 (21.1) PAP, mmHg, mean ± SD: 21.32 ± 2.43 Cardiac index, L/min/m2 of body surface area, mean ± SD: 3.20 ± 0.85 PVR, Wood units, mean ± SD: 2.09 ± 0.61 Ambrisentan (10 mg/day) n = 19 Female sex (%): 16 (84.2) Age ‐ year, mean ± SD: 58.79 ± 10.75 PAH associated with CTD (%): 19 (100.0) 6MWD, m, mean ± SD: 470.21 ±77.03 WHO FC II (%): 17 (89.5) WHO FC III (%): 2 (10.5) PAP, mmHg, mean ± SD: 19.84 ± 3.58 Cardiac index, L/min/m2 of body surface area, mean ± SD: 2.84 ± 0.64 PVR, Wood units, mean ± SD: 2.22 ± 0.93
Interventions	Intervention characteristics Placebo Timing: once daily Co‐interventions: not mentioned Compliance: low Duration of treatment period: 6 months Ambrisentan (10 mg/day) Co‐interventions: not mentioned Compliance: high Duration of treatment period: 6 months
Outcomes	Change in mean PAP Change in WHO FC Change in cardiac index Change in PVR Symptoms of SSc Quality of life (SF‐36) Lung function tests, right Heart dimensions and function NT‐proBNP Measures of disease‐related progression
Identification	Sponsorship source: GlaxoSmithKline Setting: Single‐centre, university‐based hospital Author's name: Ekkehard Grünig Institution: Centre for Pulmonary Hypertension, Thoraxklinik at Heidelberg University Hospital Email: ekkehard.gruenig@med.uni‐heidelberg.de Address: Röntgenstrasse 1, 69126 Heidelberg, Germany
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Judgement comment: "The EDITA study was a single‐center (PH Center, Thoraxklinik at Heidelberg University Hospital, Heidelberg, Germany) investigator‐initiated trial using a prospective, randomised, double‐blind (patient and investigator), parallel group, placebo‐controlled, phase IIA clinical study design. Patients were randomised 1:1 to either ambrisentan or placebo by simple randomisation." p. 3 in the study section
Allocation concealment (selection bias)	Low risk	"Placebo tablets with the same color and shape as ambrisentan were provided by GlaxoSmithKline." p. 3 in the study section
Blinding (performance bias and detection bias) All outcomes	Low risk	"Placebo tablets with the same color and shape as ambrisentan were provided by GlaxoSmithKline." p. 3 in the study section
Incomplete outcome data (attrition bias) All outcomes	High risk	Outcome data were missing in the intervention group and placebo group, and missing data were not balanced. 2/19 participants discontinued from the intervention group, and 4/19 participants discontinued from the placebo group.
Selective reporting (reporting bias)	Low risk	Reported as per protocol published at ClinicalTrials.gov.

Galiè 2003.

Study characteristics
Methods	This trial report is on a subset of participants from the BREATHE‐1 study.
Participants	Baseline characteristics Placebo n = 29 Age (mean ± SD): 44.9 ± 19.2 Female sex (%): 24 (82.8) Idiopathic PAH (%): 23 (79.3) PAH associated with CTD (%): 4 (13.8) PAP mmHg (mean ± SD): 58 ± 19 Cardiac index, L/min/m2 (mean ± SD): 2.3 ± 0.7 6MWD, m (mean ± SD): 334 ± 76 WHO FC III (%): 28 (96.6) WHO FC IV (%): 1 (3.4) Bosentan n = 56 Age (mean ± SD): 45.1 ± 15.6 Female sex (%): 48 (85.7) Idiopathic PAH (%): 48 (85.7) PAH associated with CTD (%): 7 (12.5) PAP mmHg (mean ± SD): 56 ± 15 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.8 6MWD, m (mean ± SD): 335 ± 83 WHO FC III (%): 49 (87.5) WHO FC IV (%): 7 (12.5)
Interventions	Same as BREATHE‐1
Outcomes	Cardiac index
Identification	‐
Notes	This is a subgroup analysis of BREATHE‐1 study.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	See BREATHE‐1.
Allocation concealment (selection bias)	Low risk	See BREATHE‐1.
Blinding (performance bias and detection bias) All outcomes	Low risk	See BREATHE‐1.
Incomplete outcome data (attrition bias) All outcomes	High risk	2 out of 29 participants in the placebo group and 1 out of 56 in the bosentan group discontinued the study medication. The missing data were not balanced between groups.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

MAESTRO.

Study characteristics
Methods	Study design: Randomised controlled trial Study grouping: Parallel group
Participants	Baseline characteristics Macitentan n = 114 Female sex (%): 82 (71.9) Age ‐ year (median, range): 33 (12 to 82) Down syndrome, n (%): 10 (8.8) 6MWD, m, mean (SD): 368.7 (74.5) WHO FC II (%): 69 (60.5) WHO FC III (%): 45 (39.5) SpO2 at rest, %: 84.4 ± 5.6 Mean PAP, mmHg (mean ± SD): 77.4 ± 15.1 Mean RAP, mmHg (mean ± SD): 7.9 ± 5.2 PVR, dyn/s/cm5 (mean ± SD): 1807.5 ± 792.8 PDE5 inhibitors, n (%): 35 (30.7) Simple cardiac defect, n (%): 91 (79.8) Complex cardiac defect, n (%): 23 (20.2) Placebo n = 112 Female sex (%): 68 (60.7) Age ‐ year (median, range): 31 (13 to 62) Down syndrome, n (%): 10 (8.9) 6MWD, m, mean (SD): 380.3 (76.3) WHO FC II (%): 66 (58.9) WHO FC III (%): 46 (41.1) SpO2 at rest, %: 85.2 ± 5.1 Mean PAP, mmHg (mean ± SD): 79.9 ± 17.3 Mean RAP, mmHg (mean ± SD): 7.7 ± 4.5 PVR, dyn/s/cm5 (mean ± SD): 1828.1 ± 937.4 PDE5 inhibitors, n (%): 27 (24.1) Simple cardiac defect, n (%): 80 (71.4) Complex cardiac defect, n (%): 32 (28.6)
Interventions	Intervention characteristics Macitentan Timing (e.g. frequency, duration of each episode): once daily Co‐interventions: stable background therapy for other cardiopulmonary diseases (e.g. antiarrhythmics, oral anticoagulants, digoxin, supplemental oxygen) was also allowed. Diuretics were allowed if the dose was stable for at least 1 week prior to randomisation. Iron supplementation was allowed if initiated at least 3 months prior to randomisation. Compliance: high Duration of treatment period: 16 weeks Placebo Timing (e.g. frequency, duration of each episode): once daily Co‐interventions: stable background therapy for other cardiopulmonary diseases (e.g. antiarrhythmics, oral anticoagulants, digoxin, supplemental oxygen) was also allowed. Diuretics were allowed if the dose was stable for at least 1 week prior to randomisation. Iron supplementation was allowed if initiated at least 3 months prior to randomisation. Compliance: high Duration of treatment period: 16 weeks
Outcomes	6MWD Mortality WHO FC improved WHO FC deteriorated mPAP PVR Change from baseline in 6MWD SPO2 Change in SPO2 Change in PAP Change in PVR
Identification	Sponsorship source: Actelion Pharmaceuticals Ltd. Country: International study Setting: Tertiary referral hospital Author's name: Michael A Gatzoulis Institution: The Royal Brompton Hospital Email: m.gatzoulis@rbht.nhs.uk Address: Sydney Street, London, SW3 6NP, UK
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was based on a prespecified randomisation schedule using randomisation lists generated by an independent Contract Research Organization (Almac Clinical Technologies) and their centralized randomisation system, via an Interactive Voice Response System or Interactive Web Response System."
Allocation concealment (selection bias)	Low risk	"Participants and sites remained blinded to their previous treatment allocation."
Blinding (performance bias and detection bias) All outcomes	Low risk	"Participants and sites remained blinded to their previous treatment allocation."
Incomplete outcome data (attrition bias) All outcomes	Low risk	97% participants in the macitentan group and 98% participants in the placebo group completed the study.
Selective reporting (reporting bias)	Low risk	Reported as per protocol published at ClinicalTrials.gov.

PORTICO.

Study characteristics
Methods	Study design: Randomised controlled trial Study grouping: Parallel group
Participants	Baseline characteristics Macitentan (10 mg) n = 43 Female sex (%): 21 (49%) Age ‐ year, mean (SD): 58.0 (8.7) Body mass index, kg/m2, mean (SD): 29.0 (4.8) PAH therapy: 27 (63%) 6MWD, m, mean (SD): 385.8 (100.0) PVR, dyn/s/cm5, mean (SD): 552.4 (192.8) Mean PAP, mmHg, mean (SD): 46.4 (7.9) Cardiac index, L/min/m2, mean (SD): 3.1 (0.8) Mixed venous oxygen saturation, %, mean (SD): 69.2% (9.9) Presence of oesophageal varices: 26 (60%) Presence of ascites: 12 (28%) WHO FC I: 1 (2%) WHO FC II: 27 (62·8%) WHO FC III: 15 (35%) Placebo n = 42 Female sex (%): 20 (48%) Age ‐ year, mean (SD): 59.0 (9.5) Body mass index, kg/m2, mean (SD): 29.3 (4.0) PAH therapy: 27 (64%) 6MWD at baseline, m, mean (SD): 383.2 (108.9) PVR, dyn/s/cm5, mean (SD): 521.7 (163.3) Mean PAP, mmHg, mean (SD): 43.8 (8.5) Cardiac index, L/min/m2, mean (SD): 2.9 (0.8) Mixed venous oxygen saturation, %, mean (SD): 69.9% (5.3) Presence of oesophageal varices: 28 (67%) Presence of ascites: 10 (24%) WHO FC I: 1 (2%) WHO FC II: 23 (55%) WHO FC III: 18 (43%)
Interventions	Intervention characteristics Macitentan (10 mg) Timing (e.g. frequency, duration of each episode): once daily Compliance: high Duration of treatment period: 12 weeks Placebo Timing: once daily Compliance: high Duration of treatment period: 12 weeks
Outcomes	Change in 6MWD Mortality WHO FC improved Change in mPAP Change in cardiac index Change in PVR WHO FC deteriorated
Identification	Sponsorship source: Actelion Pharmaceuticals Ltd. Country: International study Author's name: Olivier Sitbon Email: olivier.sitbon@u‐psud.fr
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Participants were randomly assigned in a 1:1 ratio, with block sizes of four, via an interactive voice and web response system (by independent contract research organisation Almac) to receive either macitentan 10 mg or matching placebo orally once a day." Judgement comment: p. 596, the first paragraph in section "randomisation and masking"
Allocation concealment (selection bias)	Low risk	"Participants were randomly assigned in a 1:1 ratio, with block sizes of four, via an interactive voice and web response system (by independent contract research organisation Almac) to receive either macitentan 10 mg or matching placebo orally once a day." Judgement comment: p. 596, the first paragraph in section "randomisation and masking"
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	9.4% participants in the macitentan group and 2.4% participants in the placebo group discontinued the study. Missing data were observed and imbalanced between groups.
Selective reporting (reporting bias)	Low risk	Judgement comment: there was no obvious selective outcome reporting based on the protocol reported in ClinicalTrials.gov.

SERAPH.

Study characteristics
Methods	Single‐centre, randomised, head‐to‐head, double‐blind trial
Participants	Baseline characteristics Sildenafil n = 14 Age, mean (range): 44.4 (28 to 62) Female sex (%): 11 (78) Idiopathic PAH (%): 12 (86) PAH associated with CTD (%): 2 (14) Cardiac index, L/min/m2 (mean ± SD): 2.3 ± 0.1 6MWD, m (mean ± SD): 290 ± 88.5 WHO FC III (%): 14 (100) Bosentan n = 12 Age, mean (range): 41.1 (27 to 55) Female sex (%): 10 (83) Idiopathic PAH (%): 11 (92) PAH associated with CTD (%): 1 (8) Cardiac index, L/min/m2 (mean ± SD): 2.2 ± 0.1 6MWD, m (mean ± SD): 304.6 ± 74.1 WHO FC III (%): 12 (100)
Interventions	Intervention group: bosentan (62.5 mg twice daily for first 4 weeks, then uptitrated to 125 mg twice daily) Timing: twice daily Co‐interventions: warfarin, diuretics, digoxin, and calcium channel blockers Compliance: high Duration of treatment period: 16 weeks Control group: sildenafil (50 mg twice daily during the first 4 weeks, then uptitrated to 50 mg 3 times daily) Timing: 3 times daily Co‐interventions: warfarin, diuretics, digoxin, and calcium channel blockers Compliance: high Duration of treatment period: 16 weeks
Outcomes	Primary outcome: change in right ventricle mass from baseline Secondary outcomes: change from baseline in 6MWD, cardiac index, Borg dyspnoea index, quality of life, and plasma B‐type natriuretic peptide level from baseline
Identification
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The medication was blinded in identical‐looking gelatin capsules and randomised using a computer generated random list by the Hammersmith Hospital pharmacy."
Allocation concealment (selection bias)	Low risk	"The medication was blinded in identical‐looking gelatin capsules and randomised using a computer generated random list by the Hammersmith Hospital pharmacy."
Blinding (performance bias and detection bias) All outcomes	Low risk	"The medication was blinded in identical‐looking gelatin capsules and randomised using a computer generated random list by the Hammersmith Hospital pharmacy."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Only 1 participant discontinued sildenafil because of death, and none discontinued bosentan during the study.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

SERAPHIN.

Study characteristics
Methods	Study design: Randomised controlled trial Study grouping: Parallel group
Participants	Baseline characteristics Placebo n = 249 Female sex (%): 184 (73.9) Age ‐ year, mean ± SD: 46.7 ± 17.03 Idiopathic PAH (%): 126 (51.0) Heritable PAH (%): 3 (1.2) PAH associated with CTD (%): 81 (32.8) PAH associated with CHD (%): 26 (10.5) PAH associated with HIV infection (%): 3 (1.2) PAH associated with drug use or toxin exposure (%): 8 (3.2) 6MWD, m, mean ± SD: 352 ± 110.6 WHO FC I (%): 0 WHO FC II (%): 129 (51.8) WHO FC III (%): 116 (46.6) WHO FC IV (%): 4 (1.6) PAP, mmHg, mean ± SD: 53.1 ± 18.1 Cardiac index, L/min/m2 of body surface area, mean ± SD: 2.44 ± 0.80 PVR, dyn/s/cm5, mean ± SD: 996 ± 784.3 Macitentan (3 mg) n = 248 Female sex (%): 187 (75.4) Age ‐ year, mean ± SD: 44.5 ± 16.26 Idiopathic PAH (%): 144 (58.3) Heritable PAH (%): 8 (3.2) PAH associated with CTD (%): 70 (28.3) PAH associated with CHD (%): 15 (6.1) PAH associated with HIV infection (%): 1 (0.4) PAH associated with drug use or toxin exposure (%): 9 (3.6) 6MWD, m, mean ± SD: 364 ± 95.5 WHO FC I (%): 0 WHO FC II (%): 138 (55.6) WHO FC III (%): 105 (42.3) WHO FC IV (%): 5 (2.0) PAH, mmHg, mean ± SD: 55.1 ± 16.7 Cardiac index, L/min/m2 of body surface area, mean ± SD: 2.36 ± 0.79 PVR, dyn/s/cm5, mean ± SD: 1044 ± 624.2 Macitentan (10 mg) n = 242 Female sex (%): 194 (80.2) Age ‐ year, mean ± SD: 45.5 ± 14.99 Idiopathic PAH (%): 134 (55.6) Heritable PAH (%): 2 (0.8) PAH associated with CTD (%): 73 (30.3) PAH associated with CHD (%): 21 (8.7) PAH associated with HIV infection (%): 6 (2.5) PAH associated with drug use or toxin exposure (%): 5 (2.1) 6MWD, m, mean ± SD: 363 ± 93.2 WHO FC I (%): 1 (0.4) WHO FC II (%): 120 (49.6) WHO FC III (%): 116 (47.9) WHO VC IV (%): 5 (2.1) PAP, mmHg, mean ± SD: 53.5 ± 17.6 Cardiac index, L/min/m2 of body surface area, mean ± SD: 2.36 ± 0.78 PVR, dyn/s/cm5, mean ± SD: 1040 ± 672.5
Interventions	Intervention characteristics Placebo Timing: once daily Co‐interventions: concomitant treatment with oral phosphodiesterase type 5 inhibitors, oral or inhaled prostanoids, calcium‐channel blockers, or l‐arginine was allowed. Compliance: high Duration of treatment period: 115 weeks (median) Macitentan (3 mg) Timing: once daily Co‐interventions: concomitant treatment with oral phosphodiesterase type 5 inhibitors, oral or inhaled prostanoids, calcium‐channel blockers, or l‐arginine was allowed. Compliance: high Duration of treatment period: 115 weeks (median) Macitentan (10 mg) Timing: once daily Co‐interventions: concomitant treatment with oral phosphodiesterase type 5 inhibitors, oral or inhaled prostanoids, calcium‐channel blockers, or l‐arginine was allowed. Compliance: high Duration of treatment period: 115 weeks (median)
Outcomes	Change in 6MWD Mortality WHO FC improved Change in mPAP Change in cardiac index Change in PVR
Identification	Sponsorship source: Actelion Pharmaceuticals Setting: Muticentre, international, university‐based hospital Author's name: Tomás Pulido Institution: Ignacio Chávez National Heart Institute Email: tpulido@prodigy.net.mx Address: Ignacio Chávez National Heart Institute, Juan Badiano 1, 4th Fl., Mexico City, 14080 Mexico
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomly assigned via an interactive voice and web response system.
Allocation concealment (selection bias)	Low risk	Confirmed with the study sponsor: a central randomisation scheme was used to assign participants, and matching placebo tablets were administered to maintain blinding.
Blinding (performance bias and detection bias) All outcomes	Low risk	"An independent clinical event committee adjudicated, in a blinded fashion"
Incomplete outcome data (attrition bias) All outcomes	High risk	59.2% participants discontinued placebo versus 45.7% participants discontinued macitentan. The missing data were not balanced.
Selective reporting (reporting bias)	Low risk	Judgement comment: clinicaltrials.gov/ct2/show/NCT00660179

STRIDE‐1.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted centrally according to a computer‐generated random number.
Participants	Baseline characteristics Placebo n = 60 Age (mean ± SD): 48 ± 14 Female sex (%): 47 (78) Idiopathic PAH (%): 37 (62) PAH associated with CTD (%): 9 (15) PAH associated with CHD (%): 14 (23) PAP mmHg (mean ± SD): 52 ± 16 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.8 PVR, dyn/s/cm5 (mean ± SD): 911 ± 504 6MWD, m (mean ± SD): 413 ± 105 WHO FC II (%): 22 (37) WHO FC III (%): 36 (60) WHO FC IV (%): 2 (3) Sitaxsentan 100 mg n = 55 Age (mean ± SD): 45 ± 14 Female sex (%): 47 (85) Idiopathic PAH (%): 23 (42) PAH associated with CTD (%): 16 (29) PAH associated with CHD (%): 16 (29) PAP mmHg (mean ± SD): 54 ± 17 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.8 PVR, dyn/s/cm5 (mean ± SD): 1026 ± 694 6MWD, m (mean ± SD): 394 ± 114 WHO FC II (%): 16 (29) WHO FC III (%): 39 (71) WHO FC IV (%): 0 (0) Sitaxsentan 300 mg n = 563 Age (mean ± SD): 44 ± 12 Female sex (%): 47 (75) Idiopathic PAH (%): 34 (54) PAH associated with CTD (%): 17 (27) PAH associated with CHD (%): 12 (19) PAP mmHg (mean ± SD): 54 ± 14 Cardiac index, L/min/m2 (mean ± SD): 2.3 ± 0.7 PVR, dyn/s/cm5 (mean ± SD): 946 ± 484 6MWD, m (mean ± SD): 387 ± 110 WHO FC II (%): 21 (33) WHO FC III (%): 42 (67) WHO FC IV (%): 0 (0)
Interventions	Intervention group: 100 mg once daily or 300 mg once daily Control group: placebo Timing: once daily Co‐interventions: warfarin, diuretics, calcium‐channel blockers, digoxin, supplemental oxygen Compliance: high Duration of treatment period: 12 weeks
Outcomes	Primary outcomes: peak oxygen consumption Secondary outcomes: 6MWD, NYHA FC, PAP, cardiac index, and PVR
Identification
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was performed centrally and stratified by centre in blocks according to a computer‐generated random number table."
Allocation concealment (selection bias)	Low risk	"Randomization was performed centrally and stratified by centre in blocks according to a computer‐generated random number table."
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	8.3% (5/60) participants discontinued placebo versus 5.9% (7/118) participants sitaxsentan. The missing data was few and balanced between groups.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

STRIDE‐2.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial. Randomisation was conducted centrally according to a computer‐generated random number.
Participants	Baseline characteristics Placebo n = 62 Age (mean ± SD): 53 ± 15 Female sex (%): 47 (76) Idiopathic PAH (%): 37 (60) PAH associated with CTD (%): 17 (27) PAH associated with CHD (%): 7 (11) PAP mmHg (mean ± SD): 49 ± 14 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.7 PVR, Wood units (mean ± SD): 11 ± 8 6MWD, m (mean ± SD): 321 ± 85 WHO FC II (%): 23 (37) WHO FC III (%): 35 (57) WHO FC IV (%): 4 (6) Sitaxsentan 50 mg n = 62 Age (mean ± SD): 57 ± 13 Female sex (%): 53 (86) Idiopathic PAH (%): 34 (55) PAH associated with CTD (%): 19 (31) PAH associated with CHD (%): 9 (14) PAP mmHg (mean ± SD): 48 ± 15 Cardiac index, L/min/m2 (mean ± SD): 2.7 ± 1.0 PVR, Wood units (mean ± SD): 10 ± 7 6MWD, m (mean ± SD): 328 ± 80 WHO FC II (%): 21 (34) WHO FC III (%): 38 (61) WHO FC IV (%): 3 (5) Sitaxsentan 100 mg n = 61 Age (mean ± SD): 55 ± 14 Female sex (%): 43 (71) Idiopathic PAH (%): 39 (64) PAH associated with CTD (%): 18 (29) PAH associated with CHD (%): 4 (7) PAP mmHg (mean ± SD): 45 ± 12 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.6 PVR, Wood units, (mean ± SD): 10 ± 7 6MWD, m (mean ± SD): 360 ± 72 WHO FC II (%): 26 (42) WHO FC III (%): 34 (56) WHO FC IV (%): 1 (2) Bosentan n = 60 Age (mean ± SD): 49 ± 16 Female sex (%): 47 (78) Idiopathic PAH (%): 34 (57) PAH associated with CTD (%): 20 (33) PAH associated with CHD (%): 6 (10) PAP mmHg (mean ± SD): 50 ± 15 Cardiac index, L/min/m2 (mean ± SD): 2.4 ± 0.6 PVR, Wood units (mean ± SD): 11 ± 5 6MWD, m (mean ± SD): 337 ± 78 WHO FC II (%): 22 (37) WHO FC III (%): 37 (62) WHO FC IV (%): 1 (2)
Interventions	Participants were randomised to receive placebo, sitaxsentan 50 mg or 100 mg orally once daily, or open‐label bosentan for 18 weeks. Timing: once daily Co‐interventions: anticoagulants, diuretics, calcium channel blockers, digoxin, and supplemental oxygen Compliance: high Duration of treatment period: 18
Outcomes	Primary outcomes: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening
Identification	‐
Notes	STRIDE‐2 trial compared sitaxsentan with both a placebo arm and an open‐label bosentan arm, therefore we have listed data for participants treated with bosentan and placebo for reference; however, we did not pool these data with the other data.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was performed centrally according to a computer‐generated random number table. Randomization was 1:1:1:1."
Allocation concealment (selection bias)	Low risk	"Randomization was performed centrally and stratified by centre in blocks according to a computer‐generated random number table."
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	17.7% (5/60) participants discontinued placebo versus 8.2% (20/245) sitaxsentan. The missing data were imbalanced between groups.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

STRIDE‐4.

Study characteristics
Methods	Multicentre, randomised, placebo‐controlled, parallel, double‐blind trial
Participants	Baseline characteristics Placebo n = 34 Age (mean ± SD): 40 ± 14 Female sex (%): 28 (82) Idiopathic PAH (%): 22 (65) PAH associated with CTD (%): 3 (9) PAH associated with CHD (%): 9 (27) PAP mmHg (mean ± SD): 64 ± 14 PVR, mmHg/L/min (mean ± SD): 15 ± 10 6MWD, m (mean ± SD): 342 ± 82 WHO FC II (%): 19 (56) WHO FC III (%): 14 (41) WHO FC IV (%): 1 (3) Sitaxsentan 50 mg n = 32 Age (mean ± SD): 44 ± 13 Female sex (%): 25 (78) Idiopathic PAH (%): 23 (72) PAH associated with CTD (%): 7 (22) PAH associated with CHD (%): 2 (6) PAP mmHg (mean ± SD): 56 ± 17 PVR, mmHg/L/min (mean ± SD): 14 ± 10 6MWD, m (mean ± SD): 350 ± 73 WHO FC II (%): 23 (72) WHO FC III (%): 9 (28) WHO FC IV (%): 0 (0) Sitaxsentan 100 mg n = 32 Age (mean ± SD): 40 ± 14 Female sex (%): 29 (91) Idiopathic PAH (%): 22 (69) PAH associated with CTD (%): 5 (16) PAH associated with CHD (%): 5 (16) PAP mmHg (mean ± SD): 63 ± 23 PVR, mmHg/L/min (mean ± SD): 14 ± 8 6MWD, m (mean ± SD): 344 ± 83 WHO FC II (%): 18 (56) WHO FC III (%): 14 (44) WHO FC IV (%): 0 (0)
Interventions	Participants were randomised to receive placebo, sitaxsentan 50 mg, or sitaxsentan 100 mg orally once daily for 18 weeks. Timing: once daily Co‐interventions: anticoagulants, diuretics, calcium channel blockers, digoxin, and supplemental oxygen Compliance: high Duration of treatment period: 18 weeks
Outcomes	Primary outcomes: change from baseline in 6MWD Secondary outcomes: change from baseline in WHO FC, Borg dyspnoea index and time to clinical worsening
Identification	‐
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was performed centrally and stratified by centre in blocks according to a computer‐generated random number table."
Allocation concealment (selection bias)	Low risk	"Randomization was performed centrally and stratified by centre in blocks according to a computer‐generated random number table."
Blinding (performance bias and detection bias) All outcomes	Low risk	Confirmed with the study sponsor: matching placebo tablets were administered to maintain blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	12.5% (4/32) participants discontinued placebo versus 10.9% (7/64) sitaxsentan. The missing data were balanced between groups.
Selective reporting (reporting bias)	Unclear risk	We did not have access to the study protocol.

6MWD: 6‐minute walk distance; BNP: B‐type natriuretic peptide; CAD: coronary artery disease; CHD: congenital heart disease; CTD: connective tissue disease; mPAP: mean pulmonary artery pressure; mRAP: mean right atrial pressure; NYHA FC: New York Heart Association functional class; PAH: pulmonary arterial hypertension; PAP: pulmonary artery pressure; PDE5: phosphodiesterase type 5; PH: pulmonary hypertension; PVR: pulmonary vascular resistance; RAP: right atrial pressure; SD: standard deviation; SF‐36: 36‐item Short Form Health Survey; SpO₂: oxygen saturation; SSc: systemic sclerosis; WHO FC: World Health Organization functional class

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Ahn 2014	Participants did not have PAH.
ASSET‐2	Participants did not have PAH.
Atsmon 2013	RCT, but populations did not have PAH.
Barst 2011	Participants were not randomised to an ERA.
Baughman 2014	Participants did not have PAH.
Benza 2007	Participants were not randomised to an ERA.
Bose 2011	Participants did not have PAH.
Bruderer 2014	Participants did not have PAH.
BUILD‐2	Participants did not have PAH.
Burgess 2008	Participants did not have PAH.
Cardenas 2020	Participants did not have PAH.
Chin 2020	Participants were not randomised to an ERA.
Coyne 2005	Participants did not have PAH.
Danaietash 2019	Participants did not have PAH.
Denault 2013	Participants did not have PAH.
Dhaun 2011	Participants did not have PAH.
Escribano‐Subias 2019	Participants did not have PAH.
Faoro 2009	Participants did not have PAH.
Fatima 2018	Participants did not have PAH.
Frey 2008	Participants did not have PAH.
Galiè 2005	Participants were not randomised to an ERA.
Galiè 2009a	Participants were not randomised to an ERA.
Galiè 2009b	Long‐term follow‐up data of RCT using sitaxsentan
Galiè 2010	Participants were not randomised to an ERA.
Ghofrani 2010	Participants were not randomised to an ERA.
Ghofrani 2017	Participants did not have PAH.
Gillies 2011	Participants did not have PAH.
Gillies 2013	Participants did not have PAH.
Givertz 2000	Participants did not have PAH.
Gomberg‐Maitland 2005	Participants were not randomised to an ERA.
Gotti 2014	Participants did not have PAH.
Gotzkowsky 2010	Participants did not have PAH.
Grander 2014	Participants did not have PAH.
Grill 2020	Participants did not have PAH.
Gutierrez 2013	Participants did not have PAH.
Han 2017	Mixed population used.
Hill 2018	Participants did not have PAH.
Hoeper 2006	Participants were not randomised to an ERA.
Howard 2019	Participants did not have PAH.
Huez 2009	Participants did not have PAH.
Hurst 2016	Participants did not have PAH.
Iversen 2010	Participants were not randomised to an ERA.
Jais 2008	Participants did not have PAH.
Kalra 2002	Participants did not have PAH.
Kaluski 2008	Participants did not have PAH.
Kefford 2010	Participants did not have PAH.
Keir 2013	Participants did not have PAH.
Keir 2014	Participants did not have PAH.
King 2008	Participants did not have PAH
Kiowski 1995	Participants did not have PAH.
Koller 2017	Participants did not have PAH.
Korn 2004	Participants did not have PAH.
Lee 2009	Participants were not randomised to an ERA.
Lindegger 2014	Participants did not have PAH.
Mazzanti 2013	Duplicate publications
McLaughlin 2003	Participants were not randomised to an ERA.
McLaughlin 2006	Participants were not randomised to an ERA.
McLaughlin 2010	Participants were not randomised to an ERA.
Mereles 2006	Participants were not randomised to an ERA.
Metersky 2011	Participants did not have PAH.
Modesti 2006	Participants did not have PAH.
Naeije 2010	Participants did not have PAH.
Nakahara 2010	Participants did not have PAH.
NCT00077584	Participants did not have PAH.
Okour 2019	Participants did not have PAH.
Oudiz 2007	Long‐term follow‐up data of RCT using sitaxsentan
Oudiz 2009	Long‐term follow‐up data of RCT using sitaxsentan
Palazzini 2010	Participants did not have PAH.
Raghu 2013	Participants did not have PAH.
Raghu 2015	Participants did not have PAH.
Reesink 2010	Participants did not have PAH.
Robbins 2006	Participants were not randomised to an ERA.
Schmetterer 1998	Participants did not have PAH.
Schmidt 2001	Participants did not have PAH.
Schuuring 2013	Participants did not have PAH.
Seheult 2009	Participants did not have PAH.
Seibold 2010	Participants did not have PAH.
Sfikakis 2007	Participants did not have PAH.
Shang 2013	Participants did not have PAH.
Shang 2016	Participants did not have PAH.
Sharma 2014	Participants did not have PAH.
Shenoy 2011	Participants did not have PAH.
Sidharta 2010	Participants did not have PAH.
Sidharta 2013a	Participants did not have PAH.
Sidharta 2013b	Participants did not have PAH.
Sidharta 2014	Participants did not have PAH.
Simonneau 2010	Participants were not randomised to an ERA.
Simonneau 2017	Participants did not have PAH.
Spence 2008	Participants did not have PAH.
Spence 2009	Participants did not have PAH.
Stavros 2010	Participants did not have PAH.
Stolz 2008	Participants did not have PAH.
Surie 2013	Participants did not have PAH.
Tanaka 2017	Participants did not have PAH.
Tanaka 2019	Participants did not have PAH.
Vachiery 2018	Participants did not have PAH.
Van Der Zander 2006	Participants did not have PAH.
Van Giersbergen 2005	Participants did not have PAH.
Webb 2017	Participants did not have PAH.
Worthington 2010	Participants were not randomised to an ERA.
Wrishko 2008	Participants did not have PAH.

ERA: endothelin receptor antagonists; PAH: pulmonary arterial hypertension; RCT: randomised controlled trial

Characteristics of studies awaiting classification [ordered by study ID]

Dwivedi 2018.

Methods	Randomisation method was not clear.
Participants	Pulmonary arterial hypertension in systemic sclerosis Inclusion criteria: Male or female participants aged 18 years Participants with systemic sclerosis PAH diagnosed as PAP > 35 mmHg NYHA FC II, III, IV SSc disease duration > 1 year Exclusion criteria: Forced vital capacity < 60% predicted Renal insufficiency Left heart disease and other relevant cardiac conditions Pregnant or breastfeeding female Participants on PAH specific therapy Liver disease
Interventions	Sildenafil 20 mg and bosentan 62.5 mg versus sildenafil 20 mg and placebo
Outcomes	Primary outcome measures: Change in pulmonary artery pressures [ Time Frame: Baseline and 6 months ] Change in pulmonary artery pressures measured by echocardiography at baseline and 6 months in participants of systemic sclerosis with PAH when treatment with single (PDE5 inhibitors at dose of 20 mg maximum up to 60 mg) versus dual therapy (PDE5 inhibitors and ERA 62.5 mg maximum up to 125 mg) for 6 months Secondary outcome measures: Change in 6MWD [ Time Frame: Baseline and 6 months ] To compare change in 6MWD at baseline and 6 months when treated with single versus dual therapy Time to clinical worsening (TTCW) [ Time Frame: Baseline and 6 months ] To compare time to clinical worsening (TTCW) in SSc participants when treated with single versus dual therapy. TTCW is defined as first occurrence of all‐cause deaths, PAH related hospitalisation, worsening of symptoms defined as a decrease of > 15% in 6MWD, and worsening of NYHA FC. Emergent side effects of sildenafil and bosentan [ Time Frame: Baseline to 6 months ] To compare the emergent side effects of sildenafil and bosentan by way of comparison of serious and non‐serious side effects
Notes

Rinaldi 2018.

Methods	Randomisation method was not clear.
Participants	PAH patients with WHO FC II or III
Interventions	Intervention characteristics Combination (ambrisentan plus tadalafil) therapy Timing: once daily Ambrisentan monotherapy Timing: once daily Tadalafil monotherapy Timing: once daily
Outcomes	Change in mPAP Change in RAP Change in PVR Change from baseline in 6MWD Change in MVO2
Notes

6MWD: 6‐minute walk distance; ERA: endothelin receptor antagonist; mPAP: mean pulmonary artery pressure; MVO₂: mixed venous oxygen saturation; NYHA FC: New York Heart Association functional class; PAH: pulmonary arterial hypertension; PAP: pulmonary artery pressure; PDE5: phosphodiesterase type 5; PVR: pulmonary vascular resistance; RAP: right atrial pressure; NT‐proBNP; N‐terminal pro B‐type natriuretic peptide; SSc: systemic sclerosis; TPR: total pulmonary resistance; WHO FC: World Health Organization functional class

Differences between protocol and review

The previous version of this review stated that studies would be excluded if the participants were not taking anticoagulants. This is no longer recommended, therefore we have removed this restriction, and included both studies of participants who were not taking anticoagulants as well as those who were.

Contributions of authors

CL: protocol initiation, study assessment, data extraction and entry, interpretation and write‐up. Overall responsibility for maintaining the review.

JC: protocol development, study assessment, data extraction, interpretation and write‐up.

YG: study assessment, data extraction, interpretation and write‐up.

BD: study assessment, interpretation and write‐up.

KL: study assessment, interpretation and write‐up.

Contributions of editorial team

Chris Cates (Coordinating Editor) checked the data entry prior to the full write‐up of the review; edited the protocol; advised on methodology; and approved the protocol prior to publication.

Hayley Barnes (Contact Editor): edited the review; advised on methodology, interpretation, and content.

Emma Dennett (Managing Editor): co‐ordinated the editorial process; advised on interpretation and content; edited the review.

Emma Jackson (Assistant Managing Editor): conducted peer review; edited the reference and other sections of the protocol and the review.

Elizabeth Stovold (Information Specialist): designed the search strategy; ran the searches; edited the Search methods section.

Sources of support

Internal sources

• All authors, Other

  The authors declare that no such funding was received for this systematic review.

External sources

• Wold Health Organization (WHO), Other

  WHO sponsored Dr Chao Liu to receive evidence‐based practice training at Monash University in Australia.

Declarations of interest

This systematic review was supported by a grant from the World Health Organization whilst CL was in residence at the Australasian Cochrane Centre. Dr Chao Liu participated in training from the Australasian Cochrane Centre (including the Cochrane Review Completion Program) whilst in Australia.

Chao Liu: none known
Junmin Chen: none known
Yanqiu Gao: none known
Bao Deng: none known
Kunshen Liu: none known

Edited (no change to conclusions)