No impact of polymorphism in the phosphodiesterase 5A gene in Cavalier King Charles Spaniels on pimobendan-induced inhibition of platelet aggregation response

1 INTRODUCTION

Pharmacogenetics describes how variation in genes may affect an individual's response to medical treatment¹ and is regarded as a fundamental step toward personalized medicine. In human medicine, it is an area of emerging interest, but limited research exists in veterinary medicine. The phosphodiesterase (PDE) 5A gene encodes a cyclic guanosine monophosphate (cGMP)-specific PDE. Platelets express both PDE5 and PDE3 and both are key enzymes involved in platelet signaling pathways ensuring the breakdown of cGMP and cAMP.^{2, 3} Polymorphisms in the PDE5A gene influence plasma cGMP concentrations under basal conditions and the response to vasodilatory stimuli in humans with heart failure.⁴ A specific polymorphism in the canine PDE5A gene (PDE5A:E90K) is associated with lower circulating cGMP in healthy dogs,⁵ possibly indicating an enhancement of PDE5A gene function and, in accordance, recent research indicates that this polymorphism decreases the response to the PDE5-inhibitor sildenafil in dogs with pulmonary hypertension.⁶

Myxomatous mitral valve disease (MMVD) is the most common cardiac disease in dogs, and Cavalier King Charles Spaniels (CKCS) are predisposed.^{7, 8} The pathophysiology of MMVD has not been fully elucidated but may be affected by altered platelet function.⁹

Standard cardiac medication regimens for MMVD in dogs target PDEs. Pimobendan is 1 of the most commonly used drugs for this condition. It is a calcium-sensitizing drug and an inhibitor of PDE3 and PDE5^{10, 11} The beneficial effect of pimobendan is mainly thought to be a result of its inotropic and vasodilatory properties.¹² In humans, where PDE inhibitors also play a role in treating cardiovascular disease, research suggests that these drugs have secondary effects on platelet aggregation and platelet inhibition.^{2, 13-20} Interestingly, a study of whole blood platelet aggregation indicated an in vitro inhibitory effect of pimobendan at high doses (10 μM) in blood samples from dogs.

Platelets express both PDE5 and PDE3, and their inhibition is thought to affect platelet signaling by decreasing the degradation of intraplatelet cyclic adenosine monophosphate (cAMP; particularly PDE3 inhibition) and cGMP (particularly PDE5 inhibition), thus leading to an increase in their concentrations. Because cAMP and cGMP are essential inhibitory intracellular second messengers, likely ensuring platelet inhibition, the net effect of PDE inhibition also might include platelet inhibition.^{2, 3, 21}

Individual differences in age of onset, progression, survival, and clinical response to treatment have been described in dogs with MMVD.^{22, 23} Likewise, large individual variations in pharmacokinetic characteristics of pimobendan have been reported (McManamey, Anna K; DeFrancesco, Terese C; Keene, Bruce W, et al. 2021. Population Pharmacokinetics of Oral Pimobendan and its Metabolite in Dogs with Myxomatous Mitral Valve Degeneration. 2021 ACVIM Forum Abstract C27).²⁴ Such variance might implicate polymorphisms in the PDE5A gene.

We aimed to investigate the presence and impact on basal platelet aggregation response of the PDE5A:E90K polymorphism in CKCS with or without preclinical MMVD. Additional objectives were to determine (1) the in vitro effect of pimobendan on adenosine diphosphate (ADP)-induced light transmission aggregometry (LTA) response at clinical treatment (0.03 μM) and high (10 μM) concentrations in platelet-rich plasma (PRP) and (2) if the potential effect of pimobendan on platelet aggregation was associated with PDE5A genotype and selected clinical variables (age, sex, severity of MMVD, hematocrit, platelet count, and echocardiographic left ventricular fractional shortening [LV FS]).

It was hypothesized that the PDE5A:E90K polymorphism would (1) be present in CKCS and increase basal platelet aggregation response because of a higher PDE5 activity and thus lower cGMP concentration and (2) attenuate the inhibitory effect of pimobendan on platelet aggregation response, because of the enhanced PDE5 activity. No association with the selected clinical variables was hypothesized.

2 MATERIALS AND METHODS

3 RESULTS

Seventy CKCS were recruited, but 18 were excluded because of macrothrombocytopenia (n = 13), serum biochemistry analysis indicating clinically relevant systemic disease (n = 2), lack of sufficient blood for analyses (n = 2), and error in sample handling (n = 1). A summary of descriptive data for the final population of 52 CKCS is provided in Table 2. All dogs were successfully genotyped and 12% of the dogs were negative wild type, 46% were positive heterozygous and 42% were positive homozygous for the PDE5A:E90K polymorphism. There were no differences in genotype among MMVD stages.

In 8 dogs, aggregation analyses were stopped after 9 minutes, and in 2 dogs, results only from run 2 were used in the analysis because of analytical errors in run 1. In 4 study dogs, ≥1 aggregation curve values fell below 0% and these were manually changed to 0 before calculation of AUC. Representative aggregation traces are given in Figure 1.

Association between basal platelet aggregation response and genotype (and additional dog characteristics and clinical variables): No association was found between genotype (or any other characteristics or clinical variable including MMVD stage, age, sex, hematocrit, and LV FS) and LTA aggregation response to ADP neither in univariable (Figure 2) nor multivariable analyses. However, a negative correlation was found between PRP platelet count and LTA aggregation response to ADP for AUC and MaxA, respectively (Figure 3). The results from univariable analyses are summarized in Table 3.

Effect of pimobendan on platelet aggregation response. Compared with the positive control (0 μM pimobendan) aggregation response, pimobendan inhibited platelet function as measured by AUC, MaxA, and Vel considerably at high concentrations (10 μM; Figure 4 and Table 4). At low concentration (0.03 μM) pimobendan only inhibited Vel (Figure 4).

Association between the inhibitory effect of pimobendan on platelet aggregation response and genotype (and additional dog characteristics and clinical variables): Genotype and platelet count in PRP were not associated with any of the LTA variables and therefore were not included in any multivariable models. The results from univariable analyses are summarized in Table 5.

In all multivariable models, only pimobendan concentration (but no other explanatory variable) was associated with platelet aggregation response. For all LTA variables, the high concentration of pimobendan (10 μM) had a higher inhibitory effect compared to the low concentration (0.03 μM). Summary of the results from multivariable analyses is given in Table 6. No association was found between platelet aggregation response variables and genotype, and the interaction between genotype and pimobendan concentration was not significant when the analyses were repeated including these explanatory variables.

4 DISCUSSION

The PDE5A:E90K polymorphism was present in CKCS with no or preclinical MMVD, without association with MMVD severity, and no association was seen between genotype and basal ADP-induced LTA response or the effect of pimobendan on platelet aggregation response. The study confirmed the previously reported inhibitory effect of pimobendan at high concentration (10 μM) on in vitro platelet aggregation but also indicated that pimobendan at clinical treatment concentration (0.03 μM) inhibited only ADP-induced aggregation Vel. Finally, platelet count in PRP showed a weak negative correlation with basal ADP-induced LTA response.

Our study confirms that findings from previous research indicating that the PDE5A:E90K polymorphism is highly prevalent in dogs also applies to CKCS.^{5, 6} Genotype-positive animals (46% heterozygous, 42% homozygous) outnumber wild-type dogs (12%). Although this polymorphism is not breed-specific, the prevalence likely reflects breed-specific selection pressures. Previous research has reported a possible functional relevance of the PDE5A:E90K polymorphism.^{5, 6} In 1 study of 15 healthy dogs, the PDE5A:E90K polymorphism was associated with a lower circulating cGMP concentration, possibly indicating enhancement of the PDE5A gene function, although cGMP protein function was not assessed.⁵ However, such an association was not identified in a subsequent study using a different study population.⁶ In theory, enhanced PDE5A gene function and increased PDE5 expression could lead to decreases in intraplatelet cGMP, which could increase basal platelet aggregation through cellular signaling pathways.²¹ However, in contrast to our hypothesis, the present polymorphism does not appear to increase basal ADP-induced LTA response or attenuate the inhibitory effect of pimobendan. Large individual variations were present in both the basal platelet aggregation response and the inhibitory response to pimobendan (Figure 4), and other genetic polymorphisms might be of importance. Furthermore, cGMP pathways are complex and may result in a variety of effects causing our simplified hypothesis not to apply.²¹ This situation might be the reason that, counterintuitively, higher plasma concentrations of cGMP are reported in dogs with congestive heart failure (CHF) compared to CKCS with preclinical MMVD.³⁴ Measures of cGMP, PDE5 or novel downstream activity measures of cAMP/cGMP such as phosphorylation of platelet vasodilator-stimulated phosphoprotein (pVASP) were not performed in our study, and represent a future direction for further elucidation of our findings.³⁵

A weak negative association between platelet count in PRP and basal LTA response (AUC and MaxA) was found. Previous research also has found associations between platelet count and platelet aggregation response in CKCS using various methodologies. Some found that low platelet count in CKCS was associated with a more hypercoagulable pattern,³⁶ others found that platelet count was positively associated with aggregation response⁹ and no association between platelet count and aggregation response also has been reported.^{37, 38} Approximately one third of CKCS have inherited macrothrombocytopenia, which usually is not associated with clinical signs of disease.^{25, 28, 39} Such dogs were included in the previously mentioned research, but excluded in our study because of the reported alterations in platelet aggregation response. Therefore, results are difficult to compare. The discrepancy between previous and present findings may be a result of the different methods of assessing platelet function or methods of performing LTA or differences in presence or means of classifying dogs as having or not having thrombocytopenia. One difference in LTA performance includes PRP standardization. Although it is possible to adjust PRP platelet count with PPP to avoid differences in platelet count, this method was not applied in our study because it is no longer recommended.²⁹ Yet, uncertainty remains about best practice when platelet count exceeds 600 × 10⁶ PLT/mL because of risk of autoaggregation. In our study, samples with PRP platelet count above 600 × 10⁶ PLT/mL (n = 15) were included because such a platelet count could reflect EDTA platelet counts within the normal range.

The reason for including the echocardiographic variable LV FS among variables with possible influence on basal aggregation and the effect of pimobendan on aggregation response is that a previous study indicated that increasing LV FS was associated with hypercoagulability based on thromboelastography in CKCS.³⁶ Our study did not find a similar association, which may be a result of different methods of aggregation analysis. In accordance with previous findings,¹² our study found an in vitro inhibitory effect of pimobendan at high concentrations (10 μM). In contrast, our study also found an inhibitory effect of pimobendan at clinical treatment concentration (0.03 μM), but only for the LTA variable Vel (and not AUC or MaxA). There are important differences between our study and the previous study¹² that might explain such a difference. First, our study used LTA, which is considered the gold standard for testing platelet function and widely used for monitoring antiplatelet treatment and is a method also applied in dogs^{40, 41} whereas whole blood aggregometry and thromboelastography were used in the previous study.¹² Second, an injectable pimobendan solution was used in our study for aggregometry, whereas a pimobendan tablet was dissolved in 50% methanol in the previous study.¹² In humans, methanol has been found to affect platelet function.⁴² Third, findings in the previous study¹² are based on 10 healthy dogs of different breeds whereas our study focused on CKCS with no or preclinical MMVD because such dogs constitute a relevant patient population for pimobendan treatment. Furthermore, the incubation period of pimobendan was 5 minutes versus 1 minute in our study. In vivo, the exposure of platelets to pimobendan would be much longer than both of these incubation periods. However, the short incubation period was chosen based on recommendations stating that longer incubation at 37°C will decrease aggregation response.⁴³ Finally, the dose considered to reflect clinical treatment concentration of pimobendan in the 2 studies differed. In the previous study,¹² 0.01 μM (3.09 ng/mL) of pimobendan was considered the mean peak plasma concentration achievable in dogs after clinical dosing protocols (0.25 mg/kg PO) based on the Vetmedin vet package insert. Yet, in our study, a mean peak plasma concentration of 0.03 μM (10 ng/mL) was used to indicate clinical relevance (Boehringer Ingelheim Animal Health, personal communication). Variation in pharmacokinetics of pimobendan in dogs are reported in previous studies with mean peak plasma concentrations ranging from 0.022 μM (7.3 ng/mL) to 0.12 μM (39.4 ng/mL) after PO administration of pimobendan at recommended dosages.^44-46 In addition, recent preliminary findings from 57 dogs with MMVD treated with pimobendan at a PO dosage of approximately 0.36 mg/kg indicated a mean peak plasma concentration of 0.095 μM or 31.8 ng/mL (McManamey, Anna K; DeFrancesco, Terese C; Keene, Bruce W et al. 2021. Population Pharmacokinetics of Oral Pimobendan and its Metabolite in Dogs with Myxomatous Mitral Valve Degeneration. 2021 ACVIM Forum Abstract C27). Thus, investigating the effect of pimobendan on platelets at concentrations in the 0.03-0.2 μM range could be relevant.

Higher concentrations (0.1 μM, 1 μM) of pimobendan were evaluated in the previous study,¹² but inhibition of platelet aggregation response only occurred at the highest concentration (10 μM). However, on the titration curves from the in-house pilot study, it appears that there may be an effect in some CKCS of pimobendan at even lower concentrations (Figure S1).

In vivo, pimobendan is rapidly metabolized in the liver to the more potent metabolite UDCG 212.^{24, 46, 47} Investigating only the effect of pimobendan in our study may not accurately reflect the substances to which canine platelets in vivo are exposed. Therefore, investigating the effect of UDCG 212 on platelet aggregation using a mixture of pimobendan and UDCG 212 or, perhaps even more relevant, blood from dogs treated with pimobendan would be relevant and could likely result in different findings than reported in our study.

Research also indicates an in vitro inhibitory effect of pimobendan on platelet aggregation in other species. In humans, platelet aggregation and thromboxane A₂ production were inhibited by pimobendan,^{17, 19} and, in cats, pimobendan also inhibited platelet aggregation.¹⁸ Furthermore, in vivo research supports this finding, suggesting an antithrombotic effect of pimobendan in humans.²⁰

Thus, our findings indicate that pimobendan exerts a considerable inhibitory effect on in vitro platelet LTA response at concentrations approximately 100 to 1000 times above clinical treatment concentration in dogs and a mild inhibitory effect on aggregation velocity at a clinical treatment concentration (0.03 μM). Pimobendan is generally well-tolerated in dogs.^{10, 32} However, in the European Medicine Agency's product summary of pimobendan characteristics it is stated that in very rare cases, signs of effects on primary hemostasis (petechiae on mucous membranes, SC hemorrhages) may be observed during treatment although a relationship with pimobendan has not been clearly established. Pimobendan is recommended for treatment of dogs with ACVIM stage B2, C, and D MMVD.⁸ In such ACVIM stages, dogs are commonly older and often suffer from comorbidities. Thus, administration of pimobendan in combination with other drugs that have antiplatelet properties or to dogs with hemostatic disorders could raise concern. Thus, the clinical importance of platelet inhibition mediated by pimobendan or the metabolite UDCG 212 and interaction with other drugs deserves further research.

Our study had some limitations. Only CKCS were included, which may limit the translational value to other dog breeds given the fact that CKCS have a different platelet aggregation response compared to other breeds.^{9, 48} On the other hand, CKCS represents a highly relevant patient population for pimobendan treatment. Different platelet counts in PRP theoretically could result in differences in platelet LTA response. As previously mentioned, PRP was not standardized to a specific platelet count because guidelines no longer recommend doing so.²⁹ Furthermore, platelet count was included in the statistical analyses and did not influence the effect of pimobendan. Also, each dog was used as its own control and only the difference between aggregation response with and without added pimobendan was analyzed. Therefore, any alteration in platelet count would affect each of the aggregation responses of each dog equally and should not have affected the comparison of different pimobendan concentrations.

5 CONCLUSIONS

The PDE5A:E90K polymorphism was present in CKCS with and without preclinical MMVD, but in contrast to our hypothesis, this polymorphism did not appear to be associated with basal LTA response. As hypothesized, pimobendan inhibited in vitro platelet aggregation response at high (10 μM) and to some extent clinical treatment (0.03 μM) concentrations. The inhibitory effect of pimobendan on in vitro platelet aggregation response was not affected by PDE5A:E90K genotype, but a weak negative correlation was found between platelet count in PRP and basal ADP-induced LTA response.

CONFLICT OF INTEREST DECLARATION

Maria J. Reimann's current affiliation is Boehringer Ingelheim Animal Health Nordics. Lisbeth H. Olsen participated in a scientific meeting in 2022 organized by Boehringer Ingelheim where accommodation was included. Joshua Stern serves as Associate Editor for the Journal of Veterinary Internal Medicine. He was not involved in review of this manuscript. No other authors declare a conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION

Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Approved by the Danish Animal Experiments Inspectorate (license no. 2016-15-0201-01074).

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.