We have performed a drug target mendelian randomization to investigate the causal effect of genetically predicted PDE5i on the risk of CHD, diabetes, prostatic cancer as well as breast cancer in the context of former epidemiological studies only detected obscure association. Strong evidence for the causal association between PDE5i and lower risk of CHD was identified. And we did find weak effect of PDE5i on lower risk of prostatic cancer and breast cancer. Nevertheless, use of PDE5i and the risk of diabetes suggested to be not significantly associated according to the results. We verified the causal hypothesis that PDE5i could significantly protect against CHD, while we did not give a concrete effect estimation to the causality considering the limitation of MR that genetic differences are life-long but typically smaller when compared to clinical interventions in magnitude of change in the exposure31.
Potential mechanism for protective effect of PDE5i on CHD
The main implication of our study that PDE5i could protect against CHD is consistent with numerous epidemiologic, invitro, animal and short-term clinical studies. Simon et al. performed an observational study of 5956 men with T2D which indicated that taking a PDE5i was significantly associated with low risk of AMI32. A retrospective observational study found that in men without CHD but who had the risk factor of CHD, PDE5i exposure was associated with reductions in major adverse cardiovascular events, and the effect was exposure degree relevant33. A cohort study including 18542 men with stable coronary artery disease (CAD) treated with PDE5i compared to alprostadil (16,548 with PDE5i and 1,994 with alprostadil) found PDE5i treatment was associated lower risk of cardiovascular mortality34.
Overall, we consider that the effect of PDE5i on lower risk of CHD may mainly depend on its cardiovascular- and cardiomyocyte-protection mechanism (Figure 3). Under the physiological condition, the bioactive NO, which produced by NO synthase (NOS) play a critical part in the cardiovascular protection through the soluble guanylate cyclase (sGC) –cGMP pathway, leading to smooth muscle relaxation and inhibition of platelet, also leukocyte adhesion3,35. PDE5i could stimulate downstream pathway of NO signal by inhibiting the effect of PDE5 breaking down cGMP into the inactive GMP6. Lifestyle factors known to protect against cardiovascular disease (CVD) as well as several commonly used cardiovascular therapies are associated with enhancive NO bioactivity36. On the other hand, In patients with high risk factors of cardiovascular diseases (diabetes, cigarette smoking, etc.) associated with endothelial dysfunction show insufficiently bioactive NO37. The reduction of bioactive NO may due to increased ROS (enhanced oxidative stress) seen in cardiovascular diseases, resulting in degradation of NO and eNOS uncoupling38. In mice models, administration of PDE5i before coronary occlusion could increase endothelial- and inducible-NO synthase39. Additionally, numerous animal and human studies have shown a continuous protective effect of PDE5i on cardiovascular system by improving hemodynamic parameters including arterial stiffness, flow-mediated dilation, and peak systolic velocity, even after discontinuation34. Given the condition that in the cardiovascular system, both Neuronal NOS (nNOS) and Endothelial NOS (eNOS) are constitutively expressed, involving in the production of bioactive NO, when the predominant eNOS becomes dysfunctional, the nNOS expressed in low level in vascular smooth muscle cells would play an auxiliary role in maintaining a certain degree of vasodilation3,40. And Silvia et al.41 using NOS inhibitor-treated rats found that even under the inhibition of NOS, the PDE5i could have a cardiomyocyte-protection shown by attenuating the accumulation of necrotic and fibrotic tissue, as well reducing myocardial lesions of rats. Simultaneously, it is worth mentioning that study found that in smokers and nonsmokers, inhibition of PDE5 by sildenafil both significantly increased NO-mediated vasodilation and the activities of PDE5 may be similar, also proving that PDE5i’s cardiovascular-protection effect still exists even under the risk factor of CHD which may lead to the inhibition of NOS42.
Moreover, treatment with PDE5i is associated with Inflammatory factor levels, including interleukin 6 which is involved in atherosclerosis progression and C-reactive protein, the most established inflammatory biomarker of CVD risk34,43. What’s more, clinical research found a significant increase of circulating endothelial progenitor cells (EPC) in young males after PDDE5 inhibitors administration, which may attribute to cGMP increase caused by PDE5-inibition triggering the upregulation of matrix metalloproteinase 9 in the BM stem cell niches, contributing to the extracellular matrix proteolysis and release of EPC in consequence44. Study found that over the days to weeks after ischemic stimulus, the EPCs could facilitate the generation of collateral circulation through an angiogenesis-mediated mechanism45. The inverse relationship between EPC counts and cardiovascular diseases has been established which also supports its vasculoprotective effect46.
Potential mechanism for protective effect of PDE5i on Breast cancer and Prostatic cancer
We have indicated a mildly significant association between PDE5i and lower risk of breast cancer as well as prostatic cancer. Many studies have reported the elevated PDE5 expression in multiple human carcinomas6. Since conflicting role of NO and cGMP in the signaling pathways involved in tumor cell proliferation and apoptosis, it is difficult to decipher the effect of PDE5 on oncogenesis and tumor progression47-49. Accordingly, whether PDE5i could be a potential anti-tumor drug is still controversial.
Breast cancer The mortality of BC in high-income countries is decreasing, however the BC incidence has been steadily increasing, though certainly caused by the advance in diagnostic methods50. Compared with normal breast tissues and benign tumors, PDE5 was found over-expressed in malignant breast tumors, resulting in decreased cGMP level as well as progression of tumor grade, stage and lymph node involvement48,51. What’s more, upregulated PDE5 was considered as relevant symbol of cancer-associated fibroblasts (CAFs) and heterotypic signaling promotion within the breast tumor microenvironment, leading to tumor progression52. Overexpression of PDE5 and its adverse effect in breast tumor highlights PDE5i’s beneficial role, especially malignant tumors.
Prostatic cancer Prostatic cancer is the most common solid malignancy found in men53. A retrospective study in 201354 found that in men with ED, using of PDE5i was associated with a reduced incidence rate of PC. However, there are more recent studies indicating that taking PDE5i is not significantly associated with the aggressiveness of PC55,56. Several established theories may explain the potential mechanism for the decreased incidence rate of PC resulted from use of PDE5i, pointing out the possible involvement of PDE5 in cell growth regulation, apoptosis of prostate cancer cell and Hippo signaling pathway53. Studies in vivo found the high expression of PDE5 in prostatic stromal compartment and different concentration of PDE5i had pronounced effect on reducing proliferation of prostate epithelial and stromal cells, respectively57. Besides, PDE-5 inhibitor could facilitate the circulatory perfusion and reduced hypoxia, since reducing hypoxia-induced PC adaptive responses and emergence of aggressive phenotype PC54.
Drug target mendelian randomization
Except randomized placebo-controlled trial, the drug target mendelian randomization would be a generally concise and easy to implement method to detect the causal association and resolve confounding and bias. With the design of conceptually random allocation of alleles at a drug target gene, drug target MR can eliminate interference and moreover, mimic lifelong effect instead of short intervention like normal RCT do. Take our study as an example, mostly in the majority of cases PDE5i are not prescribed to be used daily58, however with the design of drug target MR using genetically proxied PDE5i could mimic the daily use of PDE5i and met the criteria of continuous exposure.
Colocalization analysis
The purpose of Colocalization and MR is clearly similar, both to elucidate the nature of the relationship between traits. However, the Colocalization is generally more conservative, only aim to detect the association between two traits at a single genetic region. MR is much sterner in comparison with Colocalization. The IVs for MR are restricted to be only associated with exposure and irrelevant with downstream product of exposure or outcome. And MR normally needs be implemented in both directions, which is called bidirectional MR. The design of MR leads to a “one-way causality” from exposure to outcome whereas Colocalization indicates “association” between exposure and outcome. Since we consider Colocalization as a verification of MR. If a positive result of MR followed by proof of the existence of colocalization (H4>0.8), it would reinforce the reliability of MR. In contrast, when accompanied with result that H3 > 0.8, the result of MR may be not credible resulted from the weak IVs bias. Generally, the positive result of Colocalization provides additional support to MR result, and negative result indicates the lack of reliability of MR result.
Meaning of the finding
PDE5i was developed for its hypothesized beneficial effects on hypertension and symptomatic ischemic heart disease in 1990s. However, it did not show satisfying therapeutic effect in subsequent trials, and what makes it intriguing is its surprising effect on ED. Our results may encourage future RCT to detect the effect of PDE5i on risk of CHD, prostatic cancer and breast cancer, as well as support future change of PDE5i in clinical practice. Especially for CHD, our results enlighten that maybe future RCT should focus on the association between PDE5i use and risk of CHD in patients with or without conventional cardiac risk factors and other condition reflecting systemic endothelial damage.
Strength and Weakness
Leveraging drug target mendelian randomization, we have detected significantly causal effect of genetically proxied PDE5i use on lower risk of CHD, with colocalization analysis verifying and supporting the association. Our analytical approach is more robust to the influence of confounding and reverse causation compared with traditional observational research methods and more feasible than random control trial.
An obvious and important weakness of our study is that we merely detected significantly but non-sex specifically causal association between genetically proxied PDE5i and lower risk of CHD, and the results could be rediscussed in female only. Second weakness is the generalizability. Besides, Drug target MR can only detect effect of genetically proxied PDE5i use, which typically has smaller effects and life-long time needed to take effect compared with pharmacological agents. In other word, it would take a MR-deductively useful Drug longer time than Drugs found effective by relatively short-term RCT to show its effect on certain disease. Since we could not disclose the shortly intervening effect or exactly dose-response effect of PDE5i. Furthermore, we extracted IVs from a published though well-established MR study, making our results restricted to the reliability of the published data.