Twenty-four RCTs (3,758 participants) were included. Eleven trials (1,404 participants) assessed prostanoids. Eight trials (1,273 participants) assessed endothelin receptor antagonists. Three trials (950 participants) assessed PDE5 inhibitors. One trial (71 participants) assessed terbogrel. One trial (60 participants) assessed rosuvastatin. No placebo-controlled RCTs were identified for oxygen therapy, diuretics, warfarin and calcium channel blockers. Follow-up ranged from eight to 52 weeks; most reported at 12 or 16 weeks.
Prostanoids: Compared with conventional therapy or placebo, prostanoids statistically significantly reduced mortality (RR 0.49, 95% CI 0.29 to 0.82; 10 trials). The reduction was greater when comparing intravenous agents to placebo and when correlating the risk of mortality to the proportion of people with Class III or IV symptoms. Prostanoids were associated with statistically significant improvements in six-minute walk distance (29.4 metres, 95% CI 18.1 to 40.7; 10 trials), Borg dyspnoea score (seven trials), functional class (seven trials) and haemodynamic parameters (eight trials). Six studies reported adverse events. Prostanoids were associated with statistically significant increases in jaw pain, diarrhoea, peripheral oedema, headache and nausea.
Endothelin receptor antagonists: Compared with placebo, endothelin receptor antagonists had no statistically significant effect on mortality (eight trials). Endothelin receptor antagonists were associated with statistically significant improvements in six-minute walk distance (38 metres, 95% CI 27.2 to 48.7; seven trials), functional class (six trials), Borg score (five trials) and most haemodynamic changes (five trials). Abnormal liver function tests were reported in all studies (definitions varied). No effect was seen when all studies were combined. Studies that assessed bosentan had a statistically significant increase in abnormal liver function tests. No other adverse events were statistically significantly increased with treatment.
PDE5 inhibitors: Compared to placebo, PDE5 inhibitors had no statistically significant effect on mortality (three trials). PDE5 inhibitors were associated with statistically significant improvements in six-minute walk distance (33.7 metres, 95% CI 22.5 to 44.8; three trials) and all reported haemodynamic parameters (two trials). Borg score and functional class were each reported in only one study and were not analysed. Adverse events (which included visual disturbance, dyspepsia, flushing, headache and limb pain) were statistically significantly increased with treatment (three trials).
Other treatments: One study on terbogrel was terminated early because of adverse effects; terbogrel had no statistically significant effect on six-minute walk distance and haemodynamics. One study on rosuvastatin found no change in six-minute walk distance.