Previous studies comparing combination therapy (CT) of pulmonary vasodilators to monotherapy (MT) in patients with pulmonary arterial hypertension (PAH) report conflicting results as to whether CT is more efficacious than MT. We systematically searched the Cochrane Library, EMBASE, and MEDLINE databases for randomized controlled trials comparing CT to MT for patients with PAH. Data were pooled using the DerSimonian–Laird random-effects model. Six randomized controlled trials including 729 patients met our inclusion criteria. Follow-up ranged from 12 to 16 weeks. Compared to MT, CT resulted in a modest increase in 6-minute walk distance at the end of follow-up (weighted mean difference 25.2 m, 95% confidence interval [CI] 13.3 to 37.2). CT did not decrease mortality (risk ratio [RR] 0.42, 95% CI 0.08 to 2.25), admissions for worsening PAH (RR 0.72, 95% CI 0.36 to 1.44), or escalation of therapy (RR 0.36, 95% CI 0.09 to 1.39) and did not improve New York Heart Association functional class (RR 1.32, 95% CI 0.38 to 4.5) compared to MT. Incidence of study-drug discontinuation was similar between groups (RR 0.89, 95% CI 0.53 to 1.48). CT did not decrease the combined end point of mortality, admission for worsening PAH, lung transplantation, or escalation of PAH therapy (RR 0.42, 95% CI 0.17 to 1.04). In conclusion, this meta-analysis suggests that in PAH CT does not offer an advantage over MT apart from modestly increasing exercise capacity. However, given the paucity of good-quality data, more studies are required to define the efficacy of CT in this population before establishing final guidelines.
Pulmonary arterial hypertension (PAH) is a progressive disease of the pulmonary vasculature causing inexorable right heart failure and death. Although the condition remains incurable, the previous 15 years have seen the development of novel pharmacologic agents for treating PAH. The 3 main classes of drug currently licensed for PAH are prostanoids, endothelin receptor antagonists, and phosphodiesterase-5 inhibitors. A recent meta-analysis of monotherapy (MT) trials has suggested weak mortality benefit of vasodilator therapy over placebo. Combination therapy (CT) modulates disease pathways at multiple sites and may improve patient outcomes without necessarily increasing drug toxicity. Several randomized clinical trials of CT have been published with conflicting results in terms of efficacy (exercise capacity or clinical worsening events) but with good safety outcomes. Abraham et al published a systematic review of CT in 2010 but the study did not include searches in EMBASE and the Cochrane Library and incorporated several observational studies. Pooling of results was not performed in that study. The current “Dana Point” PAH guidelines have given a grade IIA to IIB recommendation for CT, indicating weak support for its use. In view of the inconclusive data from the published literature we performed a systematic review and meta-analysis of randomized controlled trials (RCTs) of CT for PAH and its effect on clinical worsening events and 6-minute walk distance (6MWD).
Methods
We searched MEDLINE, EMBASE, and the Cochrane Library from 1980 through January 2011. Search terms were designed to provide maximum sensitivity in detecting therapeutic trials in PAH. The search terms were “([prostanoid or epoprostenol or prostacyclin or Flolan or iloprost or Ventavis or Remodulin or treprostinil] or [‘endothelin receptor antagonist’ or bosentan or Tracleer or sitaxsentan or Thelin or ambrisentan or Volibris] or [‘phosphodiesterase 5 inhibitor’ or sildenafil or Viagra or Revatio or vardenafil or Levitra or tadalafil or Adcirca] and [pulmonary hypertension] and Humans).” There was no language restriction. We subsequently hand-searched the references of narrative reviews, guidelines, and other retrieved documents to identify any publications not identified in the database search. We excluded conference abstracts because the data therein are often preliminary and have not been thoroughly peer reviewed.
We included a study in the systematic review if (1) it was a trial in which subjects were randomly assigned to placebo or active therapy on the background of treatment with an approved PAH therapy in a parallel-group design; (2) it reported the clinical outcomes of interest; and (3) follow-up was ≥12 weeks. Two investigators (B.D.F. and A.S.) independently extracted data from each trial. Results were compared and any disagreements were resolved by consensus. Data extracted for each trial included the author, trial title, year of publication, study design, length of follow-up, number of participants and their characteristics, drug in the active treatment arm, and background PAH therapy. Pooled efficacy outcomes were 6MWD and clinical worsening end points: death, admission to a hospital for PAH deterioration, lung transplantation, and escalation of treatment (defined as addition of another approved PAH therapy or increase in epoprostenol dose >10%, where appropriate). A combined clinical worsening end point was calculated for each study as the sum of end points just described. For a surrogate safety analysis we extracted data on study discontinuations for any reason. Where available, data on pulmonary hemodynamics were qualitatively compared. When data were incompletely presented in the published article, we contacted the study sponsor directly to request the information required. For every study we scored the quality of the trial according to the scale of Jadad et al.
We used the DerSimonian–Laird random-effects model, which accounts for within-study and between-study variability to estimate pooled risk ratios (RRs) with their 95% confidence intervals (CIs) for event data. Pooled effects on 6MWD are presented as weighted mean differences with corresponding 95% CIs. Forest plots were created for each outcome. Where there were no events in 1 treatment group, we used a 0.5 continuity correction. If there were no events in either group, then any measurement of effect summarized as a ratio cannot be defined, and the trial was excluded from pooled analysis. Statistical heterogeneity was assessed using the Cochrane Q statistic (p <0.1 considered significant). We also calculated the I 2 statistics to estimate the proportion of variation attributable to between-study heterogeneity. We used R 2.11.1 with the meta package for statistical analysis. The report was drafted with reference to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.
Results
Our database search identified 5,222 studies after removal of duplicate studies ( Figure 1 ). No further articles were retrieved by manual searching. After reviewing title and abstracts to exclude irrelevant articles, 70 studies were reviewed in detail. Of these, 64 were rejected (11 narrative reviews or editorials, 34 uncontrolled case series/observational studies, 18 single-case reports). One RCT (Pulmonary Arterial Hypertension and Response to Tadalafil [PHIRST-1] study comparing tadalafil to placebo) was reported 2 times—1 time with all patients and then by a prespecified analysis of patients recruited to the study on baseline therapy with bosentan. We used the bosentan subgroup study for the meta-analysis (referred to as PHIRST-1b) and extracted data for placebo and tadalafil 40 mg/day. Overall 6 RCTs met the prespecified inclusion criteria.
Characteristics of the 6 included RCTs are presented in Table 1 . All studies were randomized and placebo controlled. Studies were double-blinded, except for the COMbination of Bosentan and aerosolized Iloprost in idiopathic pulmonary arterial hypertension (COMBI) study. All scored ≥3 on the Jadad scale, indicating good study design. In total 729 patients were enrolled; 363 received MT with placebo and 366 received CT; demographic data are listed in Table 2 . All studies presented data on clinical outcomes and 6MWD, although Bosentan: Randomized trial of Endothelin receptor Antagonists THErapy for PAH (BREATHE)-2 did not define the end points of death, admission, transplantation, and treatment escalation a priori ( Table 3 ). Raw event data are presented in Table 4 .
| Study | Year | Follow-up (weeks) | Baseline Therapy | Active Therapy Arm | Jadad Score |
|---|---|---|---|---|---|
| PHIRST-1b | 2011 | 16 | Bosentan PO | Tadalafil 40 mg/day PO | 5 (excellent) |
| TRIUMPH-1 | 2010 | 12 | Bosentan (70%) or sildenafil (30%) PO | INH treprostinil 18–54 μg 4×/day | 3 (good) |
| PACES | 2008 | 16 | Epoprostenol IV | Sildenafil 20–80 mg 3×/day PO | 5 (excellent) |
| STEP | 2006 | 12 | Bosentan PO | INH iloprost 5 μg 6–9×/day | 4 (very good) |
| COMBI | 2006 | 12 | Bosentan PO | INH iloprost 5 μg 6×/day | 3 (good) |
| BREATHE-2 | 2004 | 16 | Epoprostenol IV | Bosentan 125 mg 2×/day PO | 3 (good) |
| Placebo Arm | Active Therapy Arm | |||||
|---|---|---|---|---|---|---|
| Age, Mean ± SD | Women | IPAH/CTD/Other | Age, Mean ± SD | Women | IPAH/CTD/Other | |
| PHIRST-1b | 52±16 | 78% | 69%/18%/13% | 50 ± 13 | 79% | 52%/26%/48% |
| TRIUMPH-1 | 52 (18–75) ⁎ | 82% | 56%/31%/13% | 55 (20–75) ⁎ | 81% | 56%/35%/9% |
| PACES | 48 ± 13 | 77% | 79%/17%/4% | 48 ± 13 | 82% | 80%/16%/4% |
| STEP | 49 ± 15 | 79% | 61%/39% † | 51 ± 14 | 79% | 61%/39% † |
| COMBI | 56 ± 13 | 76% | 100%/0%/0% | 48 ± 14 | 79% | 100%/0%/0% |
| BREATHE-2 | 47 ± 19 | 55% | 91%/9%/0% | 45 ± 17 | 77% | 77%/23%/0% |
| Study | Death | Admission | Transplantation | Treatment Escalation | NYHA Change | Septostomy |
|---|---|---|---|---|---|---|
| PHIRST-1b | + | + | + | + | + | + |
| TRIUMPH-1 | + | + | + | + | + | − |
| PACES | + | + | + | + | − | − |
| STEP | + | + | + | + | + | + |
| COMBI | + | + | − | − | + | − |
| BREATHE-2 | − | − | − | − | + | − |
| Study | MT | CT | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Number | Death | Adm | Tx | Esc | DATE | Disc | Number | Death | Adm | Tx | Esc | DATE | Disc | |
| PHIRST-1b | 45 | 1 | 0 | 0 | 0 | 1 | — | 42 | 0 | 1 | 0 | 1 | 2 | — |
| TRIUMPH-1 | 120 | 1 | 5 | 0 | 0 | 6 | 9 | 115 | 0 | 4 | 0 | 0 | 4 | 13 |
| PACES | 133 | 7 | 11 | 1 | 17 | 36 | 14 | 134 | 0 | 8 | 0 | 2 | 10 | 8 |
| STEP | 33 | 0 | 4 | — | 1 | 5 | 5 | 34 | 0 | 0 | — | 0 | 0 | 4 |
| COMBI | 21 | 0 | 0 | — | — | — | 0 | 19 | 0 | 0 | — | — | — | 0 |
| BREATHE-2 | 11 | 0 | — | — | — | — | 1 | 22 | 3 | — | — | — | — | 1 |
At study enrollment there were no significant differences between treatment and placebo groups in 6MWD (5 RCTs, n = 685, weighted mean differences 3.82 m, 95% CI 6.8 to 14.4) or number of patients in New York Heart Association class III (6 RCTs, n = 714, RR 0.99, 95% CI 0.96 to 1.03). There was evidence of mild heterogeneity between studies in 6MWD (I 2 = 0%, p = 0.84) and New York Heart Association class (I 2 = 27%, p = 0.25).
In pooled analysis of change in 6MWD there was a significant increase favoring the CT group (4 RCTs, n = 450, weighted mean differences +25.2 m, 95% CI 13.3 to 37.2; Figure 2 ). There was no significant heterogeneity between studies (I 2 = 0%, p = 0.85). With regard to improvement in New York Heart Association functional class, we found no difference between MT and CT (3 RCTs, n = 183, RR 1.32, 95% CI 0.38 to 4.5), with significant heterogeneity between studies (I 2 = 77%, p = 0.01). Heterogeneity was driven by a larger number of New York Heart Association class improvements in the placebo arm of the PHIRST-1b study. For worsening New York Heart Association class, there was also no beneficial effect of CT (3 RCTs, n = 194, RR 0.78, 95% CI 0.32 to 1.89). There was no heterogeneity between studies for worsening New York Heart Association class (I 2 = 0%, p = 0.85).
In mortality analysis there was no advantage to CT compared to MT (4 RCTs, n = 624, RR 0.42, 95% CI 0.08 to 2.25; Figure 3 ). There was evidence of mild heterogeneity in mortality analysis (I 2 = 48%, p = 0.15). Number of admissions to a hospital was unchanged by CT (4 RCTs, n = 594, RR 0.72, 95% CI 0.36 to 1.44). There was mild heterogeneity between studies in hospital admission (I 2 = 21%, p = 0.28). Transplantation end points were not subjected to pooled analysis because only 1 transplantation event was reported in all 6 studies Pulmonary Arterial hypertension Combination study of Epoprostenol and Sildenafil (PACES) study, placebo arm). Examining the need for escalation of therapy, there was no significant difference between CT and MT (3 RCTs, n = 421, RR 0.36, 95% CI 0.09 to 1.39). There was no evidence of heterogeneity between studies in treatment escalation data (I 2 = 0%, p = 0.46). Premature study discontinuations were similar for CT and MT (5 RCTs, n = 642, RR 0.89, 95% CI 0.53 to 1.48). There was no between-study heterogeneity for study discontinuation (I 2 = 0%, p = 0.40). For the combined clinical worsening end point (death, admission, transplantation, treatment escalation), pooled analysis revealed that CT was not different from MT (4 RCTs, n = 656, RR 0.42, 95% CI 0.17 to 1.04; Figure 4 ). There was evidence of mild heterogeneity between studies for the end point of death, admission, transplantation, and treatment escalation (I 2 = 38%, p = 0.18). Forest plots for admission, escalations, and discontinuations are in the supplementary data file.
