1.4.1. Description of study

The GRIPHON study, a double-blind, event-driven, placebo-controlled, group-sequential trial, met the inclusion criteria and is summarized in Table 6. The primary objective of the study was to demonstrate the effect of selexipag on time to first primary outcome event in patients with PAH. The duration of the study depended on the occurrence of the primary outcome events. Patients were randomized 1:1 to selexipag or placebo via a central randomization system. Randomization was stratified by site. A block size of four was used. Identical placebo was used and the investigator and study staff, patients, monitors, and sponsor were blinded to the treatment.

1.4.2. Population

a) Inclusion and exclusion criteria

The trial included patients who had a diagnosis of PAH consistent with one of the categories of Group 1 PAH (Table 3). A few Group 1 categories were excluded, such as PAH associated with portal hypertension or schistosomiasis. Patients who were not receiving treatment for PAH and those who were receiving an ERA, a PDE5 inhibitor, or both at a dose that had been stable for at least three months were eligible for enrolment; patients who were receiving prostacyclin analogues were not eligible.

Baseline characteristics

The majority of patients were female (80%), had idiopathic PAH (56%), were classified as WHO FC II (46%) or III (53%) and had a mean time since PAH diagnosis slightly greater than two years. Eighty per cent of patients were taking one or two concomitant medications for PAH at baseline. The distribution of patients by geographical region was Eastern Europe (26.3%), Asia (19.7%), Latin America (9.5%), North America (16.7%), and Western Europe and Australia (27.8%).▬of 1,156 patients▬ were enrolled in Canada.

Prognostic risk factors were well balanced between selexipag and placebo groups at baseline.

Table 7

Summary of Baseline Characteristics in the Griphon Study.

1.4.3. Interventions

The initial dose of selexipag was 200 mcg twice daily. If this was well tolerated, the dose was increased by weekly increments of 200 mcg until 1,600 mcg twice daily was reached at week 12. If the patient experienced adverse events known to be associated with IP receptor agonists, such as headache, diarrhea, jaw pain, myalgia, flushing, and nausea, the dose could be maintained or reduced and the adjusted dose was to be defined as the maximum tolerated dose. At week 12, the maximum tolerated dose for each patient was determined, and this dose was to be kept stable for the next 14 weeks, up to the week 26 assessment of the secondary end point, change in 6MWD. After week 26, for patients with study drug dose < 1,600 mcg twice daily, investigators were allowed to further up-titrate the dose, if needed, by 200 mcg increments up to the maximum of 1,600 mcg twice daily, if the investigator identified a tolerability concern for a patient. The investigators could discontinue study treatment at their discretion.⁸

Of the selexipag-treated patients, 28% received a selexipag maintenance dose of 1,600 mcg twice daily (Table 8; i.e., the maximum selexipag dose allowed in the study) and in the placebo group, the highest number of tablets corresponding to the 1,600 mcg twice daily dose was achieved by 68% of patients. In 14 patients taking selexipag (2%), the selexipag maintenance dose was set to 0 (i.e., patients who received only the initial selexipag 200 mcg dose during the titration period and discontinued at this dose).

Table 8

Maintenance Dose of Selexipag in the Selexipag Treatment Group in Griphon.

Primary End Point

An independent CEC adjudicated all reported morbidity or mortality events. The committee was blinded to each patient’s study treatment allocation and to the occurrence of typical prostacyclin-associated adverse events. The primary outcome in GRIPHON was time to first CEC-confirmed morbidity or mortality event up to seven days after the last study drug intake. The following morbidity or mortality events were considered:

• Death (all causes)
• Hospitalization for worsening of PAH defined as any non-elective hospital stay (≥ 24 hours) for worsening of PAH

  • Worsening of PAH included signs and symptoms of right heart failure (e.g., syncope or near syncope, cyanosis, increase of breathlessness, clinically relevant deterioration of exercise capacity, decrease of oxygen saturation, increased peripheral edema, hepatomegaly, and ascites)
• Worsening of PAH resulting in need for lung transplantation or balloon atrial septostomy
• Initiation of parenteral prostanoid therapy or chronic oxygen therapy due to worsening of PAH

  • Chronic oxygen therapy was defined as a continuous use (24 hours, seven days per week) of oxygen, with the intention of maintaining the therapy long-term
• Disease progression (patients in modified WHO FC II or III at baseline) confirmed by a decrease in 6MWD from baseline (≥ 15%, confirmed by two tests on different days within two weeks) and worsening of WHO FC
• Disease progression (patients in WHO FC III or IV at baseline) confirmed by a decrease in 6MWD from baseline (≥ 15%, confirmed by two tests on different days within two weeks) and the need for additional PAH-specific therapy
• (Patients in WHO morbidity/mortality III at baseline were qualified for both the above disease progression definitions).

Secondary End Points (Listed in Order of Testing Hierarchy)

• Absolute change from baseline to week 26 in 6MWD measured at dosing trough
• Absence of worsening from baseline to week 26 in WHO FC
• Time from randomization to first CEC-confirmed death due to PAH or a CEC-confirmed hospitalization due to PAH worsening up to seven days after last study drug

  • The following two CEC-confirmed morbidity or mortality events were considered:

    -

    Hospitalization for worsening of PAH based on predefined criteria

    -

    Death due to PAH

• Time from randomization to death of all causes up to study closure
• Absolute change from baseline to week 26 in the subscale Breathlessness of Cambridge Pulmonary Hypertension Outcome Review (CAMPHOR) Symptoms. The subscale Breathlessness of CAMPHOR Symptoms was defined as the sum of the breathlessness items 11 to 18. It ranged from 0 (good) to 8 (poor)
• Absolute change from baseline to week 26 in CAMPHOR Symptoms score. The CAMPHOR Symptoms score was defined as the sum of the symptoms 1 to 25. It ranged from 0 (good) to 25 (poor).

Other End Points

• Borg dyspnea index

Safety End Points

• adverse events, serious adverse events, study drug discontinuation due to adverse events
• electrocardiogram abnormalities
• thyroid markers, bone turnover markers
• fundoscopy.

a) Primary outcome

GRIPHON had a group-sequential design for the primary outcome with options to recommend stopping for futility or for compelling efficacy at the interim analysis. Initial estimates were that 202 primary end point events would be needed for the study to have 90% power to detect a hazard ratio (HR) of 0.57 for the primary end point with selexipag, as compared with placebo, over an estimated study duration of 3.5 years, assuming a HR of 0.22 per year in the placebo group, at a one-sided type I error rate of 0.005. Authors assumed that to reach that number of primary end point events, they would need to enrol 670 patients over the course of two years, assuming an annual rate of attrition of 5%. Twenty months after the study was initiated, a blinded review of baseline data from 154 patients indicated that more patients than expected were receiving background therapy for their disease. Therefore, the hypothesized HR was changed from 0.57 to 0.65 to reflect a lower anticipated treatment effect. It was reported that to preserve the type I and type II error rates and the study duration, the required number of primary end point events was increased to 331 and the required number of patients was increased to 1,150. An independent data and safety monitoring committee performed an interim analysis, which had been planned after 202 events had occurred, with stopping rules for futility and efficacy that were based on Haybittle–Peto boundaries. The final analysis used a one-sided significance level of 0.00499.⁹

The primary end point analysis was an on-treatment analysis with follow-up data censored at the time selexipag or placebo was discontinued. No data imputation was used for the primary end point. The primary analysis of the primary outcome was performed on the full analysis set (FAS; see section 3.2.5.1 for description) using a one-sided unstratified log-rank test. Supportive time-to-event analyses of the primary outcome were performed, stratifying for variables including geographical region, PAH etiology, WHO FC at baseline and PAH medication at baseline.

Table 9

Summary of Group-Sequential Design.

b) Analysis populations

The main statistical analyses for the efficacy end points were performed on the FAS, which included all randomized patients and evaluated patients based on the group to which they were randomized. The per-protocol (PP) set included patients from the FAS, but excluded some patients who did not meet certain criteria specified in the protocol. PP set analyses were considered supplemental analyses to the main analyses. The safety analysis set included all randomized patients who had received at least one dose of the study drug.

1.7.1. Internal validity

• The GRIPHON study applied adequate methods for blinding, randomization, and allocation concealment via centralized treatment allocation procedures. Baseline prognostic factors were well balanced at the beginning of the trial between the selexipag and placebo groups. Matched placebo, including sham titration for placebo, was used to maintain blinding.
• A blinded CEC adjudicated all reported morbidity or mortality events, which would be expected to reduce bias for the analyses of the primary outcome of the study.
• There were several amendments to the protocol, including a change in the primary outcome and an increase in sample size. These were significant changes that would have potential to bias the results of the study if the reason for the changes was related to the results that were emerging at the time of the amendments. To mitigate this risk, the manufacturer performed analyses on the primary outcome that both included and excluded the 46 events that occurred prior to the sample size increase (August 16, 2011). The time-to-event results were similar when these 46 primary outcome events were included or excluded, reducing concern that the post-hoc sample size adjustments had introduced bias.
• The investigators applied means to reduce the risk of type I error through use of 99% CIs and a hierarchical testing procedure for secondary outcomes. The rationale for the order of the hierarchy was not clearly explained. The 6MWD was the first outcome in the hierarchy, but it is not as clinically relevant to patients as outcomes further down the hierarchy (e.g., time to death from any cause). Additionally, subgroup analyses were not included in the hierarchy and, given the number of comparisons tested, these are potentially subject to inflated type I error.
• Vital status was unknown in approximately 5% of patients at the end of the study. While this represents reasonably complete follow-up for a study of this duration, these missing data would have the potential to impact the analyses of rare outcomes such as death.
• Prognostic factors at baseline were well balanced, but the initiation of concomitant medications for PAH (e.g., PDE5 inhibitors and ERAs) during the study occurred in ▬ of patients in the selexipag and ▬ in the placebo groups and this may have introduced treatment bias. The direction of the bias cannot be ascertained with certainty, but if the placebo group received more aggressive concomitant therapy approaches than the selexipag patients, this could bias the results toward the placebo group. The higher rate of PAH-specific therapies in the placebo group could also have resulted in underestimating the difference in adverse event rates between selexipag and placebo because some of the adverse events occurring in the placebo group would have been caused by concomitant PAH therapies.

1.7.2. External validity

• ▬ patients in the trial were enrolled at Canadian sites. Compared with the overall group of patients in GRIPHON, patients who are candidates to receive selexipag in Canada may be slightly older, with a higher male representation. There were substantial differences in rates of concomitant PAH therapies at baseline across the different geographical regions. In spite of these potential differences and the small Canadian group in the study, the clinical expert for this review believed that the overall population of the GRIPHON study was reasonably similar to the Canadian population in which selexipag would be used.
• Selexipag is indicated for mono, dual, or triple therapy. Treatment with stable doses of ERAs and/or PDE5 inhibitors was permitted at the study start and new ERAs or PDE5 inhibitors could be added during the study. The overall trial population was heterogeneous with respect to the number and type of concomitant PAH agents and the trial was not specifically designed to compare monotherapy versus dual therapy versus triple therapy. Therefore, there is uncertainty regarding the population to which the results are generalizable, with respect to concomitant therapies.

1.8.1. Deaths

There were multiple analyses of deaths in the GRIPHON trial. A total of 100 and 105 patients in the selexipag and placebo groups, respectively, died up to study closure (HR 0.97; 99% CI, 0.68 to 1.39). Death was a component of the composite primary outcome of GRIPHON (see section 3.2.4). Death as a first primary outcome event from any cause up to seven days after the last dose of the study drug ▬ ▬in the selexipag group ▬ compared with the placebo group ▬ see first subcomponent of clinical worsening, Table 1). However, these numbers need to be interpreted cautiously because of the competing nature of the primary outcome events; i.e., the CEC-confirmed event with the earliest onset (“first” was considered. For this reason, other analyses of the deaths in GRIPHON were performed, such as analyses in Table 11. These analyses include all deaths that occurred up to the time point specified, including deaths that occurred subsequent to a primary end point morbidity event.⁶

Table 11

All-Cause and Pulmonary Arterial Hypertension–Related Deaths.

Cumulative incidence of death up to end of study is summarized in Table 12. The incidence of death up to end of study is numerically lower in the selexipag group, relative to placebo at all time points.

Table 12

Cumulative Incidence of Death up to End of Study.

1.8.2. Hospitalization

The annualized numbers of PAH hospitalizations up to the end-of-study treatment were 147 for the selexipag group and 167 for the placebo group. The annualized number of hospitalizations per year for all causes was 349 for the selexipag group and 344 for the placebo group. There were no statistically significant differences in overall hospitalization rates or number of days spent in hospital after these rates were adjusted for cumulative time on study at the group level (Table 13).

Table 13

Hospitalization for Any Cause During GRIPHON.

1.8.3. Clinical worsening

The composite primary outcome of morbidity and mortality event met the review protocol definition for clinical worsening; the results can be found in Table 1. Overall, ▬ selexipag patients ▬ patients (▬) had a primary outcome event. The HR for the primary outcome in the selexipag group versus placebo was 0.61 (99% CI, 0.46 to 0.81). The median time to clinical worsening ▬ for the selexipag group and was ▬) in the placebo group.

c) Disease progression

Thirty-two patients (5.6%) met criteria for disease progression in the selexipag group, compared with 84 patients (14.4%) in the placebo group.

Subgroup Analyses

Subpopulations of interest in this review were: 1) patients unable to achieve disease control with another PAH therapy, 2) FC, and 3) patients receiving mono or combination PAH therapy, by drug class. There were no data available for (1). Data for (2) and (3) are presented in Table 14.

Table 14

Time from Randomization to First Critical Event Committee–Confirmed Morbidity or Mortality Event up to 7 Days After Last Study Drug Intake, Full Analysis Set.

There was no evidence of interaction (P = 0.95) for study drug by PAH therapy in GRIPHON. The HRs for each concomitant therapy subgroup were similar to one another and were similar to the HR for the overall study population of 0.61 (99% CI, 0.46 to 0.81). There was no evidence of interaction for study drug by WHO FC. The HRs for the two groups (I and II versus III and IV) were similar to each other and to the HR for the overall study population.

1.8.4. WHO functional class

Absence of worsening from baseline in WHO FC at week 26 was reported for 444 of 571 (77.8%) of patients in the selexipag group and 430 of 574 (74.9%) in the placebo group (OR 1.16; 99% CI, 0.81 to 1.66), P = 0.19). Approximately 17% of selexipag patients and 20% of placebo patients had missing data for this analysis and “worsened” was imputed for these patients. Patients with FC IV at baseline were excluded as they could not shift to a worse category.

Relative to baseline, approximately ▬ of selexipag and placebo patients had improvements in FC at week 26, respectively; this analysis included patients in all FCs at baseline.⁸

1.8.5. Quality of life

The CAMPHOR questionnaires consisting of three sections, symptoms (with subscales related to energy, breathlessness, and mood), activity, and quality of life, were assigned to be completed by ▬ selexipag and ▬ placebo patients in the GRIPHON trial.⁸ Other than the data from the symptoms and breathlessness scales presented in section 3.6.8, there were no quality of life data reported for the GRIPHON trial.

1.8.6. Six-minute walking distance

The mean baseline (standard deviation [SD]) 6MWD was 358 m (76) in the selexipag group and 348 m (83) in the placebo group. Mean absolute change (SD) from baseline to week 26 in 6MWD measured at trough was -52 m (150) in the selexipag group and -66 m (148) in the placebo group. Median absolute change (range) from baseline to week 26 in 6MWD was +4.0 m (-448, 260) in the selexipag group and -9.0 m (-438, 262) in the placebo group. In the main analysis using a non-parametric ANCOVA with covariate 6MWD at baseline, the difference was ▬ (one-sided Wilcoxon-Mann-Whitney ▬). The treatment effect (location shift using Hodges-Lehmann method) versus placebo in the selexipag group was ▬). Missing 6MWD values at week 26 were imputed for 19.9% of patients in the selexipag group and 23.4% in the placebo group.⁸

1.8.7. Cardiopulmonary exercise testing

There were no outcomes reported for cardiopulmonary exercise testing.

1.8.8. Change in pulmonary hypertension symptoms

Baseline median values for the CAMPHOR Symptoms subscale were 10.0 in the selexipag group and 11.0 in the placebo group. Median absolute change from baseline to week 26 for the symptoms score was -1.0 in the selexipag group and 0.0 in the placebo group. Missing values were imputed in approximately ▬ patients. The CAMPHOR Symptoms score can range from 0 (good) to 25 (poor). The treatment effect of selexipag versus placebo was ▬⁸ The treatment effect was calculated using the Hodges-Lehmann method.

Baseline median values for the breathlessness subscale were 4.0 in both the selexipag and placebo groups. Median absolute change from baseline to week 26 in the breathlessness subscale score was 0.0 in both treatment groups. Missing values were imputed in approximately 22% of patients. The CAMPHOR breathlessness section can range from 0 (good) to 8 (poor). The treatment effect of selexipag versus placebo was 0.0 (99% CI, —0.4 to 0.0; P = 0.17). The treatment effect was calculated using the Hodges–Lehmann method.

The Borg dyspnea index rates dyspnea severity on a scale from 0 (no shortness of breath) to 10 (very, very severe shortness of breath). At baseline, median score was ▬* in both groups. At end of treatment (corresponding to individual patients’ end-of-study visit), the median score was ▬ in the selexipag group and ▬ in the placebo group.⁸

1.8.9. N-terminal prohormone of brain natriuretic peptide, mean pulmonary artery pressure, Cardiac index

The mean (SD) baseline plasma N-terminal prohormone of brain natriuretic peptide (NT-pro-BNP) was 504 ng/L (range: 13 to 11,012) in the selexipag group and 507 ng/L (range: 13 to 28,414) in the placebo group. The absolute change from baseline to end of treatment (corresponding to individual patients’ end-of-study visit) in median NT pro-BNP was 5.5 ng/L (range: —4790 to 10,873) in the selexipag group compared with 75.0 ng/L (range: —7309 to 41,586) in the placebo group. The difference in median values at the end of study was 70 ng/L.⁸

Cardiac index and mean pulmonary artery pressure were not specified as efficacy end points in the GRIPHON study.

1.9.1. Adverse events

Most patients experienced an adverse event during GRIPHON and the adverse events with at least 1% difference in incidence between selexipag and placebo are summarized in Table 15. The difference in incidence of adverse events between selexipag and placebo was 5% or greater for the following events: headache, diarrhea, pain in jaw, nausea, myalgia, vomiting, pain in extremity, and flushing (Table 15).

Table 15

Prostacyclin-Like Adverse Events in GRIPHON by Titration and Maintenance Dosing Phases.

Prostacyclin-like adverse events were separately analyzed and the difference between the selexipag and placebo groups was greater during the titration phase, compared with the difference during the maintenance phase for many of these events (headache, diarrhea, nausea, pain in jaw, myalgia, pain in extremity, vomiting, and flushing; Table 15).

1.9.2. Serious adverse events

Serious adverse events were reported in 252 patients (44%) in the selexipag group compared with 272 patients (47%) in the placebo group (Table 17). Other than PAH worsening, the most common serious adverse events were right ventricular failure, pneumonia, dyspnea, syncope and atrial fibrillation, all of which occurred at similar rates in selexipag and placebo groups.

Table 17

Adverse Events.

1.9.3. Adverse events that resulted in discontinuation of study drug regimen

Fewer patients discontinued the study drug regimen in the selexipag group (32%) compared with the placebo group (37%). This was mainly due to worsening of PAH (Table 17).

In addition to the common adverse events that resulted in discontinuation of study drug regimen listed in Table 17, adverse events of interest that led to discontinuation of the study regimen included the following: hyperthyroidism (none with placebo and one with selexipag), hypotension (two with placebo and none with selexipag), syncope (two with placebo and one with selexipag), and major bleeding event (four with placebo and two with selexipag). No events of anemia resulted in discontinuation of the study regimen.⁹

1.9.4. Notable harms

Other adverse events of interest identified in the protocol for this review that were not previously mentioned in the preceding sections included gastrointestinal adverse events, syncope, and anemia; they are listed in Table 16.

Table 16

Notable Harms of Interest.