FDA Primary Endpoints and Trial Designs in Pulmonary Arterial Hypertension Approvals

FDA endpoints in pulmonary arterial hypertension approvals: six-minute walk distance, clinical worsening and hemodynamics

FDA's pulmonary arterial hypertension (PAH) precedent falls into two broad phases. Most first-generation approvals (bosentan, ambrisentan, sildenafil, tadalafil, riociguat, inhaled and oral treprostinil, epoprostenol) rested on 12- to 16-week, placebo- or standard-therapy-controlled trials with change in six-minute walk distance (6MWD) as the primary endpoint. Since then, the major programs (macitentan, selexipag, the oral treprostinil outcome study, initial ambrisentan plus tadalafil, and the sotatercept supplement) have moved to event-driven trials with time to first morbidity/mortality or clinical-worsening event as the primary endpoint. Throughout, reviewers treated hemodynamics (PVR, cardiac output or index, mean pulmonary artery pressure) as supportive evidence of pharmacologic activity, not a substitute for clinical benefit. The two exceptions are a fixed-dose combination and a pediatric extrapolation program.

Key takeaways

• 6MWD supports an exercise-capacity claim, not a morbidity or mortality claim. Reviewers described 6MWD as a measure of exercise capacity, not a stand-in for survival or clinical events ³⁵⁵³⁵⁶, and approvals built on it were framed accordingly. For tadalafil, the review called the 40 mg dose effective "for improving exercise tolerance only" ²¹⁵.
• Walk-distance effects have often been small or contested. Remodulin's walk-distance effect was a 10 m median difference that did not reach conventional significance ⁴⁰⁸⁴⁰⁹. GRIPHON's significant 12 m week-26 effect was called of "little or no clinical meaning" by the statistical reviewer ⁵¹⁷. For sotatercept, reviewers benchmarked the 41 m effect against an estimated 33 m minimal clinically important difference ²²⁸.
• Clinical worsening as a secondary endpoint frequently failed or was contested. Tyvaso ⁴²⁷, Revatio ³⁰⁰ and Adcirca ⁵⁹⁹ did not show significant clinical-worsening benefit in their 6MWD-primary trials. For ambrisentan, FDA statisticians and clinicians disagreed on whether the secondary worsening endpoint could support labeling ⁴⁷³⁴⁷⁴.
• Event-driven composites are now the main route to outcome claims, but none established a standalone mortality benefit. Macitentan's benefit was "largely driven by worsening-of-PAH events" with no apparent effect on overall mortality ⁹³⁹⁴. Selexipag showed no demonstrated all-cause mortality benefit (HR 0.97) ⁵¹⁵.
• 6MWD lives on inside the composites. "Other worsening" or "disease progression" components typically required a confirmed ≥15% 6MWD decline plus a second criterion such as symptoms, functional class or need for new therapy ^242516654.

Phase 1: six-minute walk distance as the primary endpoint

The classic PAH design was a short, randomized, double-blind trial with change from baseline in 6MWD at week 12 or 16. Background therapy rules tightened or loosened depending on which PAH drug classes existed at the time.

Several design features recur in the reviews:

• Timing of the walk test matters. For Tyvaso, 6MWD was measured at peak exposure, 10 to 60 minutes after inhalation ⁴⁵⁸. The reviewer reported a trough difference of about 13 m, roughly two-thirds of the peak effect ⁴²⁹. For ambrisentan, reviewers qualified the benefit as shown "at least at peak drug levels" and requested a postmarketing study of walking capacity away from peak ⁴⁶⁸⁴⁶⁹.
• Blinding is scrutinized. The riociguat Division stressed blinding when interpreting walk results and called a documented 100 m improvement in an open-label study questionable ²⁷⁵. In the open-label Flolan studies, a reviewer accepted the walking-distance finding as replicated despite possible bias from permitted coaching ⁹⁹.
• Dose selection follows the 6MWD alpha. Tadalafil's 20 mg dose was described as potentially supportable as a minimum effective dose but did not meet the prespecified criterion ⁵⁹⁶.
• Trial failures are informative. Orenitram's add-on Studies 2 and 3 (16-week 6MWD trials on ERA and/or PDE5 inhibitor background) did not show an exercise benefit ⁵⁴⁷. Bosentan's WHO FC II trial used PVR and 6MWD as coprimary endpoints. Hemodynamics improved significantly, but the 19 m 6MWD advantage was not significant (p=0.08) ³⁶⁶.

Composite and responder variants of 6MWD

Two early programs did not rely on 6MWD alone:

• Ventavis (iloprost): ME97218 used a week-12 responder endpoint requiring all three of the following: at least a 10% 6MWD improvement measured 30 minutes post-dose, improvement of at least one NYHA class, and no death or PH deterioration ²¹. Response rates in WHO Group I PAH were 19% vs 4% (p=0.0033) ²². The label stated that evidence of benefit was inadequate in the CTEPH subgroup ²².
• Remodulin (subcutaneous treprostinil): The prespecified 6MWD analysis did not meet its criterion ⁴⁰. Approval rested on a post-hoc composite of walk distance and Borg dyspnea score ⁴¹⁰. FDA considered it reasonably likely to predict clinical benefit and approved under Subpart H accelerated approval (21 CFR 314.510), with a required postmarketing confirmatory study ⁶²²⁶²⁴. The indication was limited to diminishing "symptoms associated with exercise" ⁶²³. FDA later stated that approval of a 2006 efficacy supplement fulfilled those commitments ⁶²⁵. Reviewers were openly skeptical: Dr. Temple called the distance difference "very small" and questioned net benefit given infusion-site pain ³⁵.

Phase 2: event-driven morbidity/mortality trials

From macitentan onward, sponsors moved to long, event-driven trials, usually on top of background PAH therapy.

FDA's development advice for selexipag shows how the agency framed the choice. Before GRIPHON, FDA said either a 6MWT effect or a time-to-clinical-worsening effect could support approval. It cautioned that a program showing only a 6MWT effect might need dose-response data, and it accepted a single study with clinical worsening as the primary endpoint. A secondary exercise-capacity claim from that study would need to test 6MWD at the 0.01 level ⁵¹¹⁵¹². GRIPHON accordingly moved 6MWD to a week-26 secondary endpoint ⁵¹³⁵¹⁴.

How FDA has weighed 6MWD against clinical worsening

Clinical worsening as a secondary endpoint behind 6MWD

In the 6MWD-primary era, clinical worsening was typically tested lower in the hierarchy and often failed:

• Revatio: After a positive 6MWD result and significant mPAP reductions, the reviewer found no difference in time to clinical worsening, and further secondary testing stopped under the statistical plan ^299300301. A later add-on-to-epoprostenol study did show both a 26 m 6MWD effect and a significant delay in clinical worsening ³⁰²³⁰³.
• Tyvaso: Time to clinical worsening was first in the secondary hierarchy and did not differ (p=0.5829), so no further secondary endpoint could be called significant ⁴²⁷⁴²⁸.
• Adcirca: Clinical worsening improved numerically (5% at 40 mg vs 16% on placebo) but not significantly ⁵⁹⁸⁵⁹⁹. A reviewer said secondary endpoints served to support the primary endpoint, "not additional indications" ²¹⁸.
• Orenitram monotherapy: Clinical worsening (9% vs 13%) did not corroborate the 6MWD result ⁴¹⁵. Clinical reviewers recommended against approval in one review cycle after reassessing dropouts and finding the exercise result unreliable ⁴¹³⁴¹⁴.
• Bosentan: BREATHE-1 worsening through week 28 was 6.3% vs 20.3%, mostly reflecting fewer PAH hospitalizations ^359360361. Study 351 had only three events, too few for its planned time-to-event analysis ³⁶².

The ambrisentan review shows how the two disciplines can split. Clinicians favored a clinical-worsening claim. Statisticians judged that the secondary endpoint did not provide substantial evidence: results differed between trials, the analysis plan changed near study end, and type I error control was inadequate ⁴⁷³⁴⁷⁴. The riociguat Division warned that clinical-worsening endpoints can be affected by unblinding and asked for a prespecified, alpha-conserving analysis ²⁷⁶²⁷⁷.

6MWD inside the composite

In outcome trials, 6MWD rarely counts as an event on its own. SERAPHIN's "other worsening" required a confirmed ≥15% 6MWD decline, worsening symptoms, and a need for new PAH treatment, with independent adjudication ²⁴²⁹². GRIPHON's disease progression required a confirmed ≥15% decline plus worsening functional class or a need for additional therapy ⁵¹⁶. AMBITION and HYPERION used similar 6MWD-plus-another-sign constructs ⁶⁵⁴²³². As a result, clinical-worsening results are not independent of walking distance. The later Tyvaso clinical-worsening endpoint included a sustained 6MWD decrease of more than 15% ⁴³².

This shapes how reviewers read the results. For macitentan, the cross-discipline reviewer noted that the benefit was largely driven by deterioration in 6MWD and the need for new therapy ⁹³. The label described the primary benefit as primarily attributable to fewer clinical-worsening events, not as a mortality benefit ⁹⁵.

Clinical meaningfulness of the walk-distance effect

Reviews have increasingly separated statistical from clinical significance:

• Selexipag: The 12 m week-26 effect (99% CI 1 to 24) occurred while both arms deteriorated. The statistical reviewer called it of "little or no clinical meaning" ⁵¹⁷. The cross-discipline team leader still considered the small but consistent 6MWD results supportive of the primary endpoint ⁵¹⁸.
• Sotatercept: Reviewers compared the 41 m median difference with an estimated 33 m MCID and concluded the improvement was clinically meaningful ²²⁸.
• Tadalafil: The 33 m effect at 40 mg was characterized as similar to the exercise effects of many other approved PAH treatments ²¹⁴. A reviewer also noted limitations in how the test is measured ²¹³.
• Opsynvi: A review row cites placebo-corrected 6MWD improvements of about 22 to 36 m across PAH trials as context ⁵⁰¹.

Hemodynamic endpoints: supportive, rarely primary

Reviewers have consistently declined to treat PVR, cardiac output or mPAP as surrogates for clinical outcome in adults:

• A bosentan reviewer explained that PVR and cardiac output quantify the disease process and correlate with exercise changes, but had not been shown to predict morbidity or mortality ³⁶⁵.
• The riociguat Division said approved PAH products had not established a correlation between exercise and hemodynamic effects, which can appear before an exercise benefit ²⁷⁴.
• The Revatio label stated that the relationship between hemodynamic effects and 6MWD improvement was unknown ³⁰⁵.
• For sotatercept, PVR (median difference -234.6 dyn·s/cm⁵) was called a hemodynamic biomarker showing pharmacologic activity and supporting the clinical endpoints ²³⁴. Reviewers also rejected STELLAR's multicomponent responder endpoint for a Clinical Studies labeling claim because its NT-proBNP component was neither a clinical endpoint nor a surrogate ²³⁶.
• For Flolan, a reviewer said hemodynamic measurements were not efficacy endpoints, even though significant differences were reported ⁹⁹.
• For tadalafil, hemodynamics were described as generally of minor interest and analyzed descriptively ²¹⁹²²⁰.

Hemodynamics have served as the primary basis in two settings:

1. Fixed-dose combination of approved components. Opsynvi's A DUE trial used PVR at week 16 as the primary endpoint, comparing the combination against each monotherapy ³⁸⁷³⁸⁸. The combination lowered PVR by 29% vs macitentan and 28% vs tadalafil ⁵⁶⁹. An FDA meeting record states that PVR was accepted as a basis for substantial evidence that each component contributed to the effect, with the sponsor relying on prior FDA findings for tadalafil ⁵⁷⁰⁵⁷¹.
2. Pediatric extrapolation. For pediatric sildenafil, FDA accepted change in PVRI across ages 1 to 17 because younger children cannot reliably perform exercise testing. The advisory committee considered this suitable for a drug already approved in adult PAH ⁶⁷⁶. The primary peak VO2 endpoint narrowly missed (p=0.056), while combined-dose PVRI improved by -4.1 Wood units·m² (p=0.041) ⁶⁷⁸⁶⁷⁹. The bridge relied on FDA analyses of adult data linking drug-induced PVRI or mPAP changes with 6MWD changes ⁶⁸⁰⁶⁸¹. Even so, a statistical reviewer called PVRI a poor surrogate for adult 6MWD at the individual level ³⁰⁹. A 2012 review recommended against pediatric approval because of a dose-related mortality signal in the long-term extension ⁶⁸²⁶⁸³. FDA later approved a pediatric indication in January 2023 (ages 1 to 17), worded as improving exercise ability and, in children too young for standardized exercise testing, pulmonary hemodynamics thought to underlie improvements in exercise ⁶⁷⁸.

Mortality: an endpoint component, not a demonstrated claim

None of the PAH programs reviewed here established a standalone survival benefit in a controlled trial:

• Ambrisentan: deaths numerically favored the drug but not significantly. Reviewers dismissed comparisons with a historical NIH registry ⁴⁷⁵.
• Macitentan: no apparent effect on overall mortality, and PAH death alone was not significant ⁹⁴.
• Selexipag: all-cause mortality HR 0.97 ⁵¹⁵.
• Sotatercept: deaths were 7 vs 13 in ZENITH and 7 vs 6 in HYPERION. The labeling presents mortality as component or secondary information rather than a separately tested claim ^558559564.
• Bosentan: the label stated that long-term uncontrolled follow-up could not determine the effect on mortality ³⁶⁷.

Implications for PAH development programs

• Match the endpoint to the claim. A positive 12- to 16-week 6MWD trial has repeatedly supported an exercise-capacity indication ²¹⁵⁴⁶⁹. Delaying disease progression or reducing hospitalization has required an adequately powered, event-driven primary endpoint ⁵¹¹⁵¹².
• Pre-specify the hierarchy and protect alpha. Statistical reviewers have discounted secondary clinical-worsening results when the analysis plan changed late or multiplicity was not controlled ⁴⁷⁴²⁷⁷.
• Expect scrutiny of effect size, timing and blinding of 6MWD. Peak vs trough measurement ⁴²⁹, within-arm deterioration ⁵¹⁷ and open-label conduct ²⁷⁵ have all affected how reviewers read walk-distance results.
• Use hemodynamics as mechanistic support. PVR can carry a fixed-dose combination's contribution-of-components argument ⁵⁷⁰ or a pediatric extrapolation ⁶⁷⁶, but it has not substituted for clinical efficacy in a new adult PAH entity.
• Design composites with adjudicated, multi-criterion progression events. FDA reviewers have looked through the composite to its drivers (hospitalization and progression rather than death) ²⁰⁷⁹³, and labeling has reflected that ⁹⁵⁵⁶⁴.