Pulmonary Arterial Hypertension-Targeted Therapy for Pulmonary Hypertension Associated With Left Heart Disease: Bayesian and Frequentist Meta-Analysis of Randomized Controlled Trials.
Yuko Kiyohara, Taku Sakai, Lina M Freeman et al.
✦ AI-curated · Sources linked
In 30 seconds
This Bayesian and frequentist meta-analysis assessed pulmonary arterial hypertension (PAH)-targeted therapy versus placebo in patients with pulmonary hypertension associated with left heart disease (PH-LHD). It included 8 randomized controlled trials with 775 patients, evaluating endothelin receptor antagonists and phosphodiesterase-5 inhibitors. Bayesian analysis estimated a 79.9% probability of increased all-cause mortality (RR: 1.55; 95% credible interval [0.51 to 4.36]) and a 94.8% probability of increased adverse events. Frequentist analysis found no significant mortality difference but did find more adverse events with therapy (RR: 1.15; 95% CI 1.04 to 1.28).
Key findings
- Across 8 trials and 775 patients, Bayesian analysis estimated a 79.9% probability of increased mortality with PAH-targeted therapy (RR: 1.55; 95% credible interval [0.51 to 4.36]).
- Bayesian analysis estimated a 94.8% probability of increased adverse events (RR: 1.13; 95% credible interval [0.96 to 1.38]).
- Frequentist analysis found significantly more adverse events with therapy than placebo (RR: 1.15; 95% CI 1.04 to 1.28), but no significant difference in all-cause mortality.
- Subgroup analysis found no material difference between HFrEF and HFpEF; HFpEF had a 62% probability of a smaller risk increase.
Why it matters
The findings highlight uncertainty about mortality effects while indicating a consistent concern about adverse events when PAH-targeted drugs are used in PH-LHD. This distinction matters because efficacy in PAH does not establish benefit in PH-LHD, and the pooled trial evidence does not show a clear mortality advantage in this population.
What to keep in mind
The meta-analysis included only 8 trials and 775 patients; the Bayesian mortality estimate was imprecise, with a 95% credible interval spanning substantial benefit and harm.