This is the first study to investigate the effect of a vasodilator for treatment of acute pulmonary embolism in a randomized controlled trial. Our study suggests that adding epoprostenol for 24 hours to standard treatment of acute pulmonary embolism in patients with right ventricular dilatation does not result in a significant change in right ventricular overload.
Clearly, the sample size is too small to detect relatively small effects of epoprostenol. However, it should be emphasized that the point estimate of any treatment effect of epoprostenol on right ventricular diameter and function is very small: after 2.5 hours 2% in favour of placebo, after 24 hours 8% in favour of epoprostenol. These small changes are not likely to be clinically relevant. Estimation of relevant degrees of RVED or PAP reduction can possibly be derived indirectly from human thrombolysis studies, since thrombolytic treatment was associated with reduced mortality (4.7% vs 11.1%) in the setting of right ventricular dilatation in acute pulmonary embolism . Three studies reported data on RVED after thrombolytic treatment. These studies demonstrated reductions in right ventricular dimensions ranging from 15% to 49% within 24 hours [22–24]. Combining these results, a decrease in RVED of at least 15% might be the best possible estimation of clinical relevant reduction in RVED. This is twice the maximal observed effect we found in our pilot study. A post-hoc power analysis of our study data indicated that we had 80% power to detect a difference in RVED between treatment arms as large as 16%. Furthermore, it would require 55 subjects to confirm an 8% difference in treatment effect with a study of sufficient statistical power (80%). In considering such a trial, however, one would need to weigh the objective of demonstrating only a potential modest benefit in mortality reduction against the potential safety concerns and the costs of epoprostenol treatment.
Why doesn't epoprostenol have a more pronounced effect? In animal models of acute pulmonary embolism, treatment with different vasodilators, including prostacyclin, was associated with a more pronounced decrease in mean PAP of 12-33% [14, 25–27]. However, the administration of vasodilating agents in these experiments was started immediately after or even before the onset of acute pulmonary embolism. Since pulmonary vasocontriction is an early and partially transient phenomenon in acute pulmonary embolism , this might explain why animal models show a larger benefit of vasodilator treatment. In our pilot study, we treated patients with epoprostenol several hours after the onset of symptoms. The absence of a decrease in systolic PAP in either treatment group may indicate that patients were already past the phase of acute pulmonary vasocontriction. Normally, it takes time to get to the hospital, to undergo diagnosis and echocardiography, making earlier treatment difficult to realise in normal practice. This may also be the explanation for the minimal effect of the serotonin receptor blocker ketanserin on PAP and cardiac output in patients with acute pulmonary embolism . For inpatients suffering from acute pulmonary embolism, however, the situation might be different, and pulmonary vasodilatory therapy may still be effective. It might be argued that right ventricular dysfunction was not severe enough to expect a beneficial effect of epoprostenol. Indeed, TAPSE was at the lower limit of normality, suggesting at most mild right ventricular dysfunction. Pseudonormality of TAPSE can be caused by severe tricuspid valve regurgitation but this was not the case in our patients. RVFAC, however, another indicator of right ventricular function, was diminished suggesting at least moderate dysfunction. It is intriguing why TAPSE and RVFAC seem discongruent in our patients. These parameters usually correlate well, although TAPSE can be normal in patients with a low RVFAC . It is, however, possible that a pulmonary vasodilator would have a more pronounced effect in patients with more severe right ventricular dysfunction.
The choice for intravenous epoprostenol as a vasodilator merits discussion. Firstly, it could be argued that intravenous administration of epoprostenol might cause nonselective vasodilation which could lead to intrapulmonary shunting and systemic hypotension. However, hypotension or a decrease in PO2 (data not shown) did not occur in the patients who received epoprostenol infusion in our study. Inhalation of prostacyclin or nitric oxide is more selective for pulmonary vasculature and has shown to decrease pulmonary artery pressure in case reports of acute pulmonary embolism [30, 31]. However, inhalation of prostacyclin might increase dead space, possibly due to mucosal swelling  and inhalation of nitric oxide is difficult to implement in patients who are not on mechanical ventilation. Secondly, it has been suggested that intravenous administration of epoprostenol in pigs was associated with reduced right ventricular contractility , yet a recent study could not detect a change in measures of right ventricular contractility . In humans with chronic pulmonary hypertension, epoprostenol was associated with unchanged right ventricular dimensions or even increased cardiac output [35, 36]. The effect of epoprostenol on right ventricular function in acute pulmonary hypertension is not well known. Thirdly, it is possible that the dose of epoprostenol was not sufficient to decrease PAP. One might argue that uptitrating iv epoprostenol until systemic side effects like arterial hypotension occur, would be more likely to reveal beneficial effects on right ventricular overload and pulmonary pressures. However, increasing the dose of epoprostenol to 7.2 ng/kg/min, which is generally considered as high-dose, in 3 patients did not result in reduction in RVED. We did not further increase epoprostenol infusion rate because higher doses are usually associated with intolerable side-effects . Finally, it is also important to point out that other pulmonary vasodilators have different pharmacological properties, and our findings thus cannot be extrapolated to these alternatives.