This meta-analysis shows that the use of PDE5i in patients with HFrEF improves exercise capacity, cardiac performance, and pulmonary hemodynamics. The benefit of PDE5i on exercise capacity was related to the extent of PDE5i-mediated decrease in PAP. Although the RCTs of HFpEF reported heterogeneous results, the effects of PDE5i in patients with HFpEF were consistent with the interpolated relationship between PAP and outcome parameters. Our findings suggest the therapeutic potential of PDE5i in patients with HF, in relation with pulmonary hemodynamics.
PDE5i treatment is beneficial in HFrEF patients
In this meta-analysis, we observed significant and consistent benefits conferred by PDE5i treatment for patients with HFrEF. These findings are in line with the results of previous meta-analysis by
Zhuang et al. [
4], but we further extended the previous study and provided more comprehensive results on the changes in exercise capacity and hemodynamic parameters by including the most recently published RCTs. Therefore, we successfully showed the therapeutic effect of PDE5i in HF patients in relation with the baseline pulmonary hemodynamics.
More importantly, we summarized the occurrence of clinical events including death and hospitalization from the previous RCTs. Our results also suggest that PDE5i would reduce the occurrence of death and hospitalization in patients with HFrEF, whereas it does not increase adverse events or affect BP and HR. The interpretation of this result needs caution, because the number of RCTs that reported clinical outcomes was small, and the follow-up duration of each trial was not longer than 12 months. However, this finding can be a meaningful result, when interpreted in relation to the benefits by PDE5i on the other study outcomes: the improved cardiac performance and pulmonary hemodynamics might result in the better prognosis.
Potential benefits of PDE5i in HF have been suggested from previous animal studies: PDE5i prevented myocardial dysfunction by anti-remodeling, anti-apoptotic and anti-inflammatory effects in various disease models of HF [
28‐
30]. Our group developed a rat model of chronic myocardial regurgitation, and showed that the use of sildenafil significantly attenuated LV remodeling and prevented exercise intolerance, probably through the anti-apoptotic and anti-inflammatory effects [
28]. Cardioprotective effect of sildenafil was supported in a chronic model of doxorubicin cardiotoxicity [
31], as well as an ischemic cardiomyopathy model [
32]. Another PDE5i, udenafil, also demonstrated cardioprotective effect in rats exposed to pressure-overload cardiac hypertrophy: udenafil prevented cardiac remodeling and improved exercise capacity and survival, through inhibition of fibrosis and apoptosis, and modulation of inflammatory cytokines in the hypertrophied myocardium [
29]. These findings suggest that the long-term use of PDE5i in HF might be beneficial for myocardial reverse remodeling, improvement in LV function, and probably prevention of cardiovascular mortality. Until now, there has been no study to suggest the prognostic benefit of PDE5i in HFrEF patients, and our meta-analysis may serve as a hypothesis-generating study for future trials [
1,
33].
Of note, among the 15 included RCTs, there was one RCT of udenafil for HFrEF that showed significant benefits on cardiac function and exercise capacity [
9]. Together with the hemodynamic benefit and safety of udenafil reported in previous studies [
29,
34‐
36], the use of udenafil could be a promising therapeutic measure in patients with HFrEF, when compared to sildenafil. Given the concerns on the differences in pharmacodynamic profile between the PDE5i medications and the potential adverse events, further trials are required for the use of other PDE5i, such as tadalafil, vardenafil, and avanafil, in HF patients.
Inconclusive effect of PDE5i in HFpEF patients
The action mechanism of PDE5i in HF was suggested to be relaxation of pulmonary vessels and reverse remodeling effects on pulmonary vasculature and LV myocardium [
37].
Guazzi M et al. reported the first RCT on the effect of PDE5i in HFpEF patients, and demonstrated significant improvement in LVEF and reduction in pulmonary resistance [
12]. However, the following two RCTs of PDE5i in patients with HFpEF reported no significant benefits. The RELAX trial showed that the use of sildenafil in HFpEF patients was not associated with improved exercise capacity and cardiac performance [
13]. Similarly, in a recent RCT by
Hoendermis et al., sildenafil did not reduce pulmonary pressure or improve cardiac performance in patients with HFpEF [
15].
Previous trials reported inconsistent results and the pooled effects are inconclusive in determining the effect of PDE5i in HFpEF. We also showed that the pooled effect of PDE5i needs to be interpreted with caution, considering the potential influence from the RCT by
Guazzi M et al. [
12] on the outcome measures. However, considering the pathophysiologic background of PH-LHD as well as the potential effect of PDE5i in this context, we attempted further analysis of the pooled results according to pulmonary hemodynamics.
Pulmonary hemodynamics modulates the effect of PDE5i in HF patients
In HF patients, diastolic dysfunction leads to passive backward transmission of filling pressures, which results in PH. This condition is defined as Ipc-PH [
3]. Among patients with HF and PH, some patients develop further increase in pulmonary pressures, defined as Cpc-PH, due to the combination of pulmonary vasoconstriction, decreased NO availability, increased endothelin expression, desensitization to natriuretic peptide-induced vasodilation, and vascular remodeling [
2,
38]. The stage of Ipc-PH in HF patients indicates that the cause of PH is pure mechanical component, other than endothelial dysfunction or limited NO bioavailability [
3]. By contrast, Cpc-PH is a more advanced stage in which NO availability and remodeling of pulmonary vasculature and LV have more important roles [
39].
Our meta-regression analysis showed that the PDE5i-mediated PAP decrease had a linear interpolation relationship with the improvement in exercise capacity, and that the higher baseline PAP is an important determinant of the PDE5i-mediated benefits. Results of our sensitivity analysis also supported the significant association between the PDE5i-mediated improvements in PAP and exercise capacity. These findings are in accordance with the early study by
Lewis GD et al., which showed direct correlations between baseline pulmonary resistance and PDE5i-mediated improvement in exercise capacity [
5]. These results also suggest that the benefits of PDE5i in HF patients would be related to the degree of PH. Add to this, our data suggest that the use of PDE5i in HFpEF patients would be beneficial for those with higher baseline pulmonary pressure and whose pulmonary pressure could be reduced by PDE5i, namely, those with Cpc-PH. In particular, the beneficial effect of PDE5i in HFpEF reported by
Guazzi M et al. may be attributable to the more advanced PH at baseline [
12]. In the RELAX trial, however, the pulmonary pressure was not specified as an entry criterion, and participants had a mean of PASP 41 mmHg and calculated mPAP of 27 mmHg [
13,
40]. Therefore, the lack of a beneficial effect of PDE5i in the RELAX trial may be associated with the relatively low pulmonary pressures and the small proportion of participants with Cpc-PH. Similarly, there were only six patients (12%) with Cpc-PH in the recent RCT by
Hoendermis et al., which reported no beneficial effect of PDE5i [
15].
Therefore, the heterogeneous results reported in the RCTs of HFpEF patients should not be regarded as the failure of PDE5i in this patient population, but rather that the data might be interpreted in consideration of the baseline PH stage of the study participants. In the present study, there was a significant relationship between the PDE5i-mediated PAP decrease and the baseline pulmonary pressure. This finding suggests that the suboptimal or nonsignificant effects of PDE5i in the RCTs of HFpEF could be related to the presence of Ipc-PH, but not attributable to the preserved LVEF. As an integration of previous trials, our meta-analysis suggests that the effect of PDE5i might be dependent to the pulmonary hemodynamics in HF patients, and that the patients with Cpc-PH could be benefited by PDE5i therapy. Moreover, our findings are in accordance with pathophysiologic background and recent concepts of PH-LHD: the Cpc-PH indicates a more advanced form of PH-LHD, and the increased PAP in Cpc-PH patients can be alleviated by PDE5i. Given the potential hemodynamic benefit conferred by PDE5i in Cpc-PH, we believe that further trials would provide more important relevance in the management of PH-LHD.
Limitations
This study has several limitations in addition to the inherent methodological limitations of meta-analysis. First, the main study outcomes of our meta-analysis were not reported in all of the included RCTs, and cardiac catheterization was not performed in some RCTs. In our meta-analysis, the PASP values were converted to mPAP values, which might not be exactly proportional to PAP measurement by cardiac catheterization [
21]. The lack of those outcome parameters limited the overall statistical power of this meta-analysis. Second, there is a possibility that the patients included in the prior study were duplicated in the following studies. Although we tried to minimize any potential influence from the duplicated patients, the interpretation and generalization of our findings need caution. Third, the study population of the RELAX trial was significantly larger than that of other trials; therefore, the results of HFpEF analyses might have been influenced by the RELAX trial results. Moreover, among the RCTs of HFpEF, there was only one study that reported positive findings favoring the use of PDE5i (
Guazzi M et al. Circulation 2011) [
12], whereas the other RCTs of HFpEF demonstrated negative results. Nevertheless, the inconsistent results of the HFpEF analysis can be explained by the interpolated relationship between pulmonary hemodynamics and PDE5i therapy, which showed overall concordance among the included RCTs of both HFrEF and HFpEF. Despite the potential influences from the larger study population of the RELAX trial and its limitation of not using pulmonary hemodynamics as an entry criterion, the integrated results support the therapeutic effect of PDE5i in HF patients with advanced PH. Also, our sensitivity analysis showed that the associations between the PDE5i-mediated potential benefits and pulmonary hemodynamics were not influenced from the RELAX trial or the RCT by
Guazzi M et al., suggesting that our findings would have clinical implications. Fourth, the concept of PH-LHD with differentiation between Ipc-PH and Cpc-PH was not an inclusion criterion in the selected RCTs. Because the study participants of each RCT may have had different hemodynamic profiles, the simple interpolation analysis could not be applied to each individual patient. However, we showed the overall association between pulmonary hemodynamics and the effects of PDE5i, suggesting the potential application of PDE5i in PH-LHD. In addition, our findings imply that future trials on HFpEF need to consider the pulmonary hemodynamics not only as study endpoints but also as important eligibility criteria.