Comparative Study Of Prognosis Improvement From Precision Antiplatelet Therapy And Artificial Neural Network In Patients After PCI

PART ONE:The Benefits of Precision Antiplatelet Therapy Based on CYP2C19 Genotype in Patients Undergoing Percutaneous Coronary Intervention:A Retrospective Cohort Study BACKGROUNDCoronary artery disease(CAD)is one of the most serious illnesses affecting human health with an increasing morbidity.Compared with simple pharmacotherapy,percutaneous coronary intervention(PCI)can reduce the risk of death and recurrent myocardial infarction(MI),and has become a more adopted treatment method for CAD.However,it has been found that within 18 months after PCI,the incidence of stent thrombosis(ST)is about 1.5%,which is a thorny problem of interventional therapy.The combination of aspirin and P2Y12 inhibitors is the cornerstone of postoperative therapy and plays an important role in the inhibition of platelet reactivity and thrombotic events.Clopidogrel as the most widely used P2Y12 receptor inhibitor can effectively reduce ST,recurrent MI and other adverse cardiovascular events.However,there are about 4%-30%of patients treated with clopidogrel,showing weakened antiplatelet efficacy or ineffective and even thrombotic events.This phenomenon is known as high on-treatment platelet reactivity(HTPR).A large number of studies have confirmed that HTPR is a major risk factor for ischemic events after PCI.Thus,in order to improve the antiplatelet effect of clopidogrel,personalized antiplatelet therapy(PAPT)was developed and is becoming more important in clinical areas.In recent years,the relationship among polymorphism of CYP2C19,efficacy of clopidogrel and clinical prognosis after PCI has been studied,but the results are not consistent.OBJECTIVESIn this study,a non-randomized retrospective approach was adopted to further evaluate the benefits of genotype detection based PAPT in CAD patients after PCI.Compared with previous studies,this study was designed to compare the benefits obtained from CYP2C19*2&*3 genotype detection and experiential antiplatelet therapy in patients after coronary stenting.METHODSPatients undergoing coronary stent placement between March 2012 and August 2015 after pretreatment with P2Y12 receptor antagonist(clopidogrel or ticagrelor)and aspirin(100 mg/day for at least 5 days or loading dose of 300 mg)were eligible for this single-center non-randomized retrospective study.Patients in the genetic detection group received an antiplatelet therapy according to CYP2C19*2&*3 phenotype.CYP2C19*1/*1 carriers,also called extensive metabolizers(EMs),received a loading dose of 300 mg clopidogrel before PCI and 75 mg/day as a maintenance dose for at least 1 year after PCI;CYP2C19*1/*2 or*1/*3 genotype carriers,also called intermediate metabolizers(IMs),received a loading dose of 300 mg clopidogrel before PCI and 150 mg/day as a maintenance dose for at least 1 year after PCI;CYP2C19*2/*2,*2/*3,or*3/*3 genotype carriers,also called poor metabolizers(PMs),received a loading dose of 180 mg ticagrelor before PCI and 90 mg twice daily as a maintenance dose for at least 1 year after PCI.Patients in the conventional therapy group,without CYP2C19*2&*3 genotype detection,received clopidogrel(300 mg loading dose before PCI + 75 mg/day as a maintenance dose)or ticagrelor(180 mg loading dose before PCI + 90 mg twice daily as a maintenance dose)for at least 1 year after PCI,based on clinical experience.Generally,doctors prefer to double the dosage of clopidogel or change to ticagrelor in case of patients undergoing emergency PCI,as well as left main,bifurcation and more branches lesion coronary stenting,while they prefer to choose conventional dose of clopidogel in case of patients with high risk of hemorrhage such as patients with old age,renal insufficiency or with higher CRUSADE score.Patients were interviewed after 1 year(±15 days)during a clinical examination,after re-hospitalization,or through telephone call.The primary end point was a composite of major adverse cardiac events(MACEs),including death from any cause,MI and stroke,for the 1 year period after PCI.Secondary end points included the individual components of the primary end point,hemorrhagic events and ST.STATA 12.0 was used for all statistical analyses.p<0.05(2-sided)was considered statistically significant.RESULTSIn all,there were 984 patients(460 in genetic detection group and 524 in conventional therapy group)entered this retrospective study.No significant differences were found between the 2 groups regarding the clinical characteristics such as age,height,body mass index,blood pressure and other indexes.Moreover,199(43.26%)patients with EMs(CYP2C19*1/*1),208(45.22%)patients with IMs(182 with CYP2C19*1/*2(39.57%)and 26 with CYP2C19*1/*3(5.65%)),and 53(11.52%)with PMs(30 with CYP2C19*2/*2(6.52%),5 with CYP2C19*3/*3(1.09%),and 18 with CYP2C19*2/*3(3.91%))were found in the genetic detection group.The procedural characteristics including target vessel and drug elution were not different between these two groups.Primary end point incidence was 3.70%in the genetic detection group and 5.53%in the conventional therapy group(p = 0.173).Secondary end points incidence including death,MI,stroke,ST,and bleeding events were also not significantly lower in the genetic detection group compared with the conventional therapy group.CONCLUSIONSIn the genetic detection group,the proportion of PMs is lower than EMs and IMs.The benefits of CYP2C19*2&*3 genotype guided PAPT in CAD patients after stenting is not superior to experiential PAPT.PART TWO:Prediction of Clinical Prognosis in Patients Undergoing Percutaneous Coronary Intervention Based on Artificial Neural Network Model BACKGROUND The current study found that certain clinical factors will increase the risk of cardiovascular adverse events and bleeding events in patients after PCI.The long-term prognosis and its influencing factors after PCI have become a research hotspot in recent years.For the prognosis prediction of disease,the commonly used method is Logistic regression.However,Logistic regression analysis has very strict requirements on the sample size,which depends on the characteristics of the model and data.When the sample size is small,the probability estimate and the interpretation of confidence interval must be very cautious,and when there is a high degree of collinearity between the independent variables,or if the variation of the dependent variable is too small,a larger sample size is required.Artificial neural network(ANN)principle can be used to analyze the multiple factors and the relationships between the factors which have unclear relationships,and it is a system based on the imitation of human brain structure and function.ANN,due to their excellent ability of non-linear mapping,generalization,self-organization and self-learning,have been proved to be of widespread utility in digital signal processing,system modeling,automatic control,and others.The applications of neural networks have received a great deal of attentions in clinical medicine.By properly choosing neural networks structures and training the weights,researchers could use neural networks to perform medical outcome prediction.Neural networks have the capacity to "learn" how to make a diagnosis through the information presented to them.However,at present,the use of ANN model for prognosis after PCI has not been reported in the literature.OBJECTIVESIn this study,the retrospective cohort data of the first part of this thesis was applied to the Neural Network Toolbox of MATLAB R2014a software,and the ANN model was established by backpropagation(BP)algorithm to predict the clinical prognosis of patients after PCI.The accuracy of the predictive model was evaluated,and the value of CYP2C19 genotype detection in the model was evaluated too.METHODSPatients' information and follow-up data are the same as the first part of this thesis.The patients' information includes:age,gender,whether to receive CYP2C19 genetic testing,systolic blood pressure,diastolic blood pressure,body mass index,whether it is the first interventional treatment,the number of stent(s)in each branch of the coronary artery,low density lipoprotein cholesterol,total cholesterol,hemoglobin,platelet count,P2Y12 receptor antagonist,statin,whether to use tirofiban,whether to take warfarin,whether to use bivalirudin.The follow-up endpoint included death from any cause,myocardial infarction,stroke,bleeding events and stent thrombosis.The follow-up period is 1 yearThe ANN used in this experiment was a one-hidden-layer perception with the improved conjugate grads BP algorithm that had 29 neurons in the input layer,29 in the hidden layer,and 1 in the output layer.Where in and Nn(n varies from 1 to 29)are the nth input neuron and the nth hidden-layer neuron,respectively,and Ok is the output neuron for kth set of input data.These artificial neurons passed information from one layer to the other,in a process that imitated a human synapse,and the processed information produced an output signal.The number of interactions or learning cycles of the network was established upon trial and error in all the interactions,and used the reduction of the cross entropy as learning criterion.To evaluate the value of CYP2C19 genotyping in the ANN predictive model:remove the CYP2C19 genotype test in the input layer and model it with the same network structure and parameters.After modeling,two models were used to predict the same data,and the prediction results were tested by chi-square test.P<0.05 was considered statistically significant.RESULTSAfter data cleansing,the information of 892 patient were included in the data set,and the data was randomly divided into two groups using MATLAB R2014a software:modeling group(n=800)and validation group(n=92).The modeling model was used to construct the ANN model.The prediction accuracy rate was 80.0%.The validation group data was used to evaluate the value of CYP2C19 genotype detection in the ANN prediction model.It was found that the inclusion of this variable was not statistically significant for the construction of the model(p= 0.3972).CONCLUSIONS1.Based on the patients' age,gender,whether to receive CYP2C19 gene test,blood pressure,body mass index,whether it is the first intervention,position and quantity of stent(s),low density lipoprotein cholesterol,total cholesterol,hemoglobin,platelet count,anticoagulation and lipid-lowering drugs,ANN model can be constructed and with high accuracy for prognosis prediction within 1 year after PCI.2.Whether CYP2C19 genotype test is performed has no obvious significance for the construction of ANN model.PART THREE:The Benefits of Precision Antiplatelet Therapy in Patients Undergoing Percutaneous Coronary Intervention:A Meta-analysisBACKGROUNDDual antiplatelet therapy consists of P2Y12 receptor antagonist such as clopidogrel,prasugrel,or ticagrelor,in combination with aspirin.This therapy represents the main medical treatment in patients with acute coronary syndrome(ACS)after PCI,and in secondary prevention of atherothrombotic events.Among them,clopidogrel used to be the most broadly prescribed P2Y12 receptor inhibitor with undisputable benefits especially in combination with aspirin.However,since 2003,studies suggested that the pharmacodynamic effect of clopidogrel considerably varies among individuals,implying that it may lead to the occurrence of ischemic or bleeding events.These events were once known as clopidogrel resistance,or clopidogrel non-responsiveness,and they are now identified as HTPR.Up to 25%-50%of patients treated with clopidogrel show inadequate pharmacological response and a consequent inadequate protection from major adverse cardiac events(MACEs).Thus,in order to improve the antiplatelet effect of clopidogrel,PAPT is becoming more and more important.At present,several platelet function test(PFT)methods have been well established,including light transmission aggregometry(LTA),multiple electrode aggregometry(MEA),vasodilator stimulated phosphoprotein(VASP)phosphorylation,VerifyNow,plateletworks,platelet function analyzer(PFA-100),thromboelastography(TEG),cone and platelet analyzer(CPA).Nevertheless,the routine measurement of platelet reactivity has not been widely implemented,and lack of consensus concerning optimal method and the best cut-off value associated with clinical risk has hindered the consideration of platelet function testing in clinical guidelines.Another approach is the genotype test.As a pro-drug,clopidogrel requires enteric and hepatic transformation by the cytochrome P450(CYP)system to exert its antiplatelet effect.After absorption,up to 85%of clopidogrel is hydrolyzed by carboxyesterase-1 to an inactive metabolite.The remaining amount is metabolized to the active compound.CYP2C19 enzyme seems to have the most prominent role in the production of clopidogrel active metabolite,while CYP2B6,CYP1A2,CYP3A/A5,and CYP2C9 show lesser involvement.So far there were many researches evaluated the association between CYP2C19 loss-of-function alleles and clinical efficacy of clopidogrel.However,the conclusions of these studies were not consistent.OBJECTIVESIn this study,a meta-analysis was performed to further evaluate the benefits of PAPT in CAD patients.Compared with previous studies,our meta-analysis included all PAPT available,not only PFTs but also genetic detection as intervention.METHODSAll published RCTs enrolling CAD patients treated with PAPT according to genetic detection or PFT for at least 1 month were selected.Controlled intervention was the standard antiplatelet therapy not guided by genetic detection or PFT.The search involved various computerized databases:PubMed(up to 31 May 2017),Embase(up to 31 May 2017),and Cochrane Controlled Trials Register(up to May 2017),searching the following items:(genotype OR(genetic testing)OR(genetic polymorphism)OR(platelet function testing)OR(platelet reactivity)OR(VerifyNow)OR plateletworks OR(light transmission aggregometry)OR(multiple electrode aggregometry)OR(Platelet Function Analyzer)OR(vasodilator stimulated phosphoprotein)OR(thrombelastography)OR(Cone and Platelet Analyzer))AND((clopidogrel)OR(cangrelor)OR(elinogrel)OR(prasugrel)OR(ticagrelor)).Trials belonging to the following categories were excluded:(i)non-RCTs;(ii)subjects not treated with P2Y12 receptor antagonist;(iii)treatment duration<1 month;and(iv)trials with no mention of MACEs or bleeding events prevention.Results of the outcome were expressed as radio ratio(RR)with 95%confidence interval(CI)for each study.A pooled effect was calculated using a random-effects model.Heterogeneity was assessed using Q and I2 statistic.Subgroup analysis and meta-regression were performed to localize the source of heterogeneity.Sensitivity analysis was performed through the trim and fill method.Publication bias was evaluated using funnel plot and Egger's regression method.All statistical analyses were performed using Review Manager 5.3 and STATA 12.0.Statistical significance was defined asp<0.05(2-sided).RESULTSA total number of 1078 relevant articles were retrieved from PubMed(561),EMBASE(633),and Cochrane Controlled Trials Register(854).Among them,14 studies reporting the differences in terms of MACEs and bleeding events between 9497 patients with and without the PAPT were considered eligible for our meta-analysis(4878 randomized to personalized antiplatelet therapy and 4619 to control).The studies showed a significantly increased risk of MACEs(RR 0.58,95%CI 0.42?0.80,p=0.001),ST(RR 0.60,95%CI 0.41?0.87,p = 0.008)and MI(RR 0.43,95%CI 0.21?0.88,p = 0.02)in patients receiving PAPT compared to the standard therapy group.Furthermore,no statistically significant difference was observed between the above two groups in cardiovascular(CV)death(RR 0.77,95%CI 0.51?1.16,p = 0.21),bleeding events(RR 0.96,95%CI 0.81?1.13,p = 0.59)and ischemic stroke(RR 0.81;95%CI 0.39?1.66,p= 0.57).However,the heterogeneity was substantial in MACEs(Chi2 = 58.71,p<0.00001,I2=81%)and MI(Chi2 ?35.57,p<0.0001,I2 =78%)groups.In order to identify the heterogeneity source in MACEs group,meta-regression and subgroup analyses were performed.The results revealed that the benefits of receiving PAPT had no clear linear relation with the follow-up period(Coef.? 0.120,Std.Err = 0.059,p = 0.068).However according to subgroup analysis,there was obvious difference among the benefits after 1,6 and 12 months(among subgroups:Chi2 = 8.90,p = 0.01,I2 = 77.5%),but the heterogeneity was not well located.Another subgroup analysis revealed that the benefits of PAPT was more significant in the HTPR subgroup(RR 0.46;95%CI 0.27?0.80,p = 0.006),but not in the no mention subgroup(RR 0.70;95%Cl 0.48-1.00,p = 0.05).However,the heterogeneity was substantial in each subgroup(HTPR:Chi2 = 13.29,p=0.02,I2 ?62%;No mention:Chi2 =29.29,p<0.0001,I2=83%).Finally,based on PAPT strategy difference,the source of heterogeneity was located.Although the numbers of trials in each subgroup were small,the heterogeneities in each subgroup were not obvious(LTA:Chi2 = 0.63,p = 0.43,I2=0%;MEA:Chi2=1.36,p=0.51,I2=0%;VerifyNow:Chi2=0.96,p= 0.62,P 0%;VASP:Chi2=7.04,p=0.03,P=72%),and it was found that all methods could obtain obvious benefits except VerifyNow and VASP(LTA:RR 0.55,95%CI 0.37?0.83,p=0.004;MEA:RR 0.47,95%CI 0.32?0.67,p<0.0001;CYP2C19 gene test:RR 0.29,95%CI0.14?0.64,p0.002;VerifyNow:RR 1.08,95%CI0.98?1.19,p=0.12;VASP:RR 0.15,95%CI 0.02?1.00,p=0.05).Sensitivity analysis was performed through the Trim and Fill method,and no obvious difference was found after Trim and Fill processes,suggesting that the pooled estimates in each group were relatively robust.According to the funnel plot and Egger's regression,obvious publication biases were found in MACEs(Intercept:-2.726,95%CI-3.657?-1.796,p=0.000),CV death(Intercept:-1.741,95%CI-2.472?-1.010,p=0.000),MI(Intercept:-1.649,95%CI-3.090?-0.207,p=0.030),and ST(Intercept:-1.582,95%CI-2.801?0.362,p= 0.016)groups.CONCLUSIONSIn conclusion,in patients undergoing coronary stenting,PAPT could reduce the risk of MACEs,ST and MI.However,there was no significant increase in CV death,bleeding events and ischemic stroke.The preventive effects of PAPT on MACEs were more evident during a short follow-up period and in patients with HTPR.All methods could provide benefits except VerifyNow and VASP,but no statistical difference was found between all strategies results.