【作者】 黄盛文；

【导师】 徐湘民；

【作者基本信息】 南方医科大学 ， 细胞生物学， 2008， 博士

【摘要】 背景与目的华法林是一种双香豆素衍生物,自20世纪40年代发明以来,由于其有效的抗凝作用和低廉的价格,至今仍是临床上使用最多的口服抗凝药物。但华法林的有效血药浓度范围狭窄,剂量偏小会降低治疗血栓形成的效果,剂量偏大则会增大出血的风险,甚至引发大出血等副作用。应用华法林抗凝治疗的同时,也成倍地增加了患者出血的发生率,甚至危及生命,尤其是在治疗初期的数周到数月内。使用华法林的另一个特点是其稳定剂量在不同种族间及个体间存在着较大差异,不同个体间稳定剂量的差异可达20倍以上。因此,将不同患者的华法林剂量调整到既安全又有效的范围,是长期以来临床用药上一个非常棘手的问题。造成华法林用量个体差异的原因很多,可分为非遗传因素和遗传因素。非遗传因素主要有年龄、性别、体表面积、药物的相互作用、饮食习惯和疾病状态等。然而,非遗传因素的影响程度较为有限,并非华法林用量个体差异的主要原因。近年来,随着药物基因组学的进展和华法林药理作用分子机制的阐明,遗传因素在华法林用量个体差异中的作用越来越受到人们的重视。目前,已知与华法林的药效学和药动学相关的基因达30余种,其中维生素K环氧化物还原酶复合体亚单位1基因(VKORC1)的基因多态性和细胞色素P450 2C9基因(CYP2C9)是影响华法林用量个体差异最主要的两个遗传因素。维生素K环氧化物还原酶(VKOR)是华法林作用的靶蛋白。华法林通过抑制VKOR,使无活性的氧化型(还氧化物型)VK无法还原为有活性的还原型(氢醌型)VK而起到抗凝作用。多个VKORC1的SNP位点被发现与华法林用量个体差异相关,其中常见的两个SNP是第一内含子的1173C＞T和3′UTR的3730G＞A。这些SNP位点呈高度连锁不平衡,经单倍型分析分为A、B两组单体型,A组(H1和H2)与华法林低剂量相关,B组(H7、H8和H9)与高剂量相关。CYP2C9是华法林最主要的代谢酶。到目前为止,已发现的CYP2C9的等位基因有30种,其中以*1(野生型)、*2(Arg144Cys)、*3(Ile359Leu)最为常见。携带有变异性等位基因的患者,其华法林代谢酶的活性明显低于野生型,并且,其出血的危险性增加2～3倍。最近已有学者对这两个基因在华法林个体差异中的贡献大小进行了研究,这些研究显示VKORC1和CYP2C9的多态性对华法林用量个体差异的贡献比例分别6-37%和5-22%。近年来,一些针对特定人群,结合遗传因素和非遗传因素的华法林稳定剂量预测算法相继出现。但这些算法在广泛应用于临床之前,必须进行前瞻性的随机病例-对照研究,以评价其安全性和剂量预测的可靠性。目前,针对华法林的基于药物基因组学的大样本前瞻性研究仅见于欧美国家白种人的报道,由于存在人种间的差异,有必要对其他种族人群特别是中国人进行研究。与我们的研究几乎同时进行的只针对病例组的中国人群前瞻性研究报告最近刚发表(2008年1月),我们在通过大样本回顾性研究建立华法林稳定剂量测算法的基础上,进一步设计了中国人群的病例组和对照组的前瞻性临床和实验研究。本研究的目的是以中国人群作为研究对象,明确VKORC1和CYP2C9多态性与中国人华法林用量个体间差异的相关性,结合非遗传因素,在回顾性研究的基础上建立稳定剂量预测算法;经前瞻性随机病例-对照研究评价该算法的实用性和可行性,探寻适合中国人群的华法林个体化用药模式。此外,本研究还建立了一种基于荧光染料SYBR GreenⅠ的real-time PCR SNP位点的基因分型方法,使其能够达到简单、快速、准确和经济的要求,适用于普通临床实验室开展基因分型。设计与方法本研究内容包括三部分内容:(1)候选基因的选取,基因分型方法的建立和评价;(2)基于药物基因组学的华法林个体化用药的回顾性研究及稳定剂量预测算法的建立;(3)基于药物基因组学的华法林剂量预测算法指导临床用药的前瞻性随机病例对照研究。针对VKORC1 T6484C和CYP2C9 A1075C,两个SNP位点,我们分别设计了基于DHPLC和基于荧光染料SYBR Green-Ⅰ的real-time PCR两种基因分型方法。以测序结果作为金标准,对这两种方法的准确性、所用时间和成本进行比较,选择较为快速、准确、成本低廉的方法作用于本研究的基因分型。在回顾性研究中,共266例已达稳定剂量的临床病人进入研究计划,我们详细记录了患者的年龄、性别、身高、体重、吸烟、饮酒、同时服用的药物和饮食习惯等。每例采集血样2ml做基因分析。将各因素分别与稳定剂量作相关性分析,以α=0.05为检验水准,剔除无统计学意义的因素,确定与华法林用量相关的因素。将有统计学意义的因素与稳定剂量作多元回归分析,得出决定系数R~2,即各因素的总和能解释华法林用量个体差异原因的比例。据此建立多元回归方程作为华法林稳定剂量的预测算法。在前瞻性研究中,收集符合入选标准的同期初次服用华法林的病例112例,随机分为实验组和对照组病。实验组病例服药前先测定VKORC1和CYP2C9的基因型,代入华法林剂量预测算法计算出稳定剂量,患者的前3次用药按此剂量服药,再以凝血酶原时间(prothrombin time,PT)的国际标准化比值(internationalnormalized ratio,INR)的变化逐步调整至实际稳定剂量;对照组按传统方式用药,并逐步调整至稳定剂量。INR的监测频率为:从用药开始到出院期间为每天1次;出院后为每周1次,获得稳定剂量后为每月1次。详细记录到达稳定剂量的天数,出现副作用的情况及时间。实验数据用生存分析法处理。终点事件定为:(1)到达稳定剂量的时间(天);(2)到达出现副反应(INR＞3.5,出血或静脉栓塞)的时间(天)。用log-rank检验比较实验组与对照组终点事件的差异;用Cox比例风险回归模型分析各个因素对终点事件的影响以及实验组和对照组患者获得稳定剂量时间的风险比HR(hazard ration)。结果与讨论本研究建立了两种基于SNP位点的基因分型方法,即分别采用DHPLC和real-time PCR对VKORC1和CYP2C9进行基因分型的技术。两种方法都具有很高的准确性,DHPLC方法的检测结果与测序结果的符合率为99%,而real-timePCR方法的检测结果与测序结果完全一致。与DHPLC相比,real-time PCR方法更简便、省时和廉价,而且具有一定的通量,适合前瞻性研究中对VKORC1和CYP2C9基因型的快速检测。在回顾性研究中的266例患者中,华法林稳定剂量的范围是0.625～8.125mg/d,平均2.95±1.18mg/d,变异系数(CV)40%,最大剂量和最小剂量之间相差13倍。这进一步证实人群中华法林稳定剂量确实存在很大的差异。在影响华法林用量个体差异的非遗传因素中,我们选取性别、年龄、体表面积、吸烟、饮酒、是否伴有糖尿病、主要的治疗疾病和平均INR等8种因素进行分析。将这8种非遗传因素分别与华法林稳定剂量(经自然对数转换)进行单因素相关分析,显示性别、吸烟、饮酒、是否伴有糖尿病、主要的治疗疾病5种等级因素与华法林稳定剂量没有显著性相关关系,而年龄、体表面积和平均INR均与华法林稳定剂量具有显著的线性关系。患者华法林的稳定剂量随着年龄的增加而降低,在不考虑其他因素的情况下,从10到90岁,年龄每增加10岁,华法林稳定剂量降低0.2mg,年龄对华法林用量个体差异的贡献是7.1%。同样,在不考虑其他因素的情况下,体表面积每增加1m~2,华法林稳定剂量增加1.54mg,体表面积对华法林用量个体差异的贡献是4.9%。经单因素相关性分析,VKORC1和CYP2C9两个基因均与华法林稳定剂量具有显著性相关关系。在不考虑其他因素的情况下,VKORC1和CYP2C9对华法林用量个体差异的贡献分别是26.5%和21%。把单因素相关分析中与华法林稳定剂量有显著相关性的因素采用逐步回归分析的方法,最后在多元回归中对华法林用量个体差异有显著影响的因素有年龄、体表面积、VKORC1和CYP2C9。根据最佳回归模型得到的多元回归方程即华法林稳定剂量的预测算法,该方程可解释华法林稳定剂量个体差异54.1%的原因。在前瞻性研究中,共有112例患者入选,最后完成整个随访过程的病例有95例,其中实验组49例,对照组46例。实验组和对照组患者获得稳定剂量的平均时间分别是27.4±1.8天和34.8±1.9天,中位时间分别是24.0±1.6天和33.0±4.4天。两组患者获得稳定剂量的时间有显著性差异(P=0.011),说明本研究中的稳定剂量预测算法可有效地减少患者华法林稳定剂量的调整时间。采用Cox比例风险回归模型对组别、性别、年龄、体表面积、VKORC1和CYP2C9基因型进行分析,结果显示对华法林稳定剂量调整时间有统计学意义的因素只有组别和年龄(P=0.019,P=0.037)两个因素。在随访期内,在相同病例数的情况下,实验组获得稳定剂量的患者数是对照组的1.764倍(HR:1.764,95%CI:1.084-2.869,P=0.022)。患者获得稳定剂量的时间有随着年龄增加而增加的趋势,在4个年龄组中,≥50岁年龄组与其他年龄组比较,获得稳定剂量的时间显著增加。在前瞻性研究的4个次要观察指标中,只有获得稳定剂量的比例在实验组和对照组中有显著性差异,分别是81.6%和63%。其他3个指标,即住院调整时间,副反应发生率和患者到达副反应的时间在两组患者中均没有显著性差异。实验组和对照组共有69例患者在随访期间获得稳定剂量,这69例患者的平均预测剂量和平均实际剂量分别为2.89±0.66 mg/d和2.82±1.09mg/d,预测剂量比实际剂量平均高0.07±0.81mg/d。预测剂量与实际剂量之间具有显著的线性关系(r=0.676,P＜0.001)。有66.7%的患者的预测剂量与实际剂量相比在可接受范围内。本研究说明结合遗传因素和非遗传因素的华法林稳定剂量预测算法可以很好地指导临床用药。结论回顾性研究表明年龄、体表面积,以及VKORC1和CYP2C9多态性对华法林用量个体差异具有显著的影响,其中遗传因素在华法林用量个体差异中起着主导作用。结合遗传因素和非遗传因素建立的华法林稳定剂量预测算法,可解释华法林用量个体差异54.1%的原因。前瞻性随机病例对照研究结果显示实验组获得稳定剂量的调整时间显著少于按传统用药的对照组,患者的预测稳定剂量与实际稳定剂量之间有很高的相关性。前瞻性研究证实我们提出的华法林预测算法可以很好地指导临床用药,有助于提高华法林应用的安全性和有效性。更多还原

【Abstract】 Context and objectivesAs a derivant of bishydroxycoumarin,warfarin is the most commomly prescribed oral anticogulant in the world from 40s 20th centry when it was invented.For the narrow therapeutic index,it is easy to lead to adverse effects,such as thrombogenesis when the dose is insufficient or bleeding when the dose is excessive.The risk of bleeding is increased several times,and even cause death when use warfarin as anticogulant,especially in the first weeks to months during the early stages of treatment.Another characteristic of warfarin therapy is the large different stable dose between different races as well as interindividuals,with more than 20-fold of the stable dose range.As such,it is difficult for for clinicians to safely and efficiently prescribe warfarin for diffenent patients.Multiple factors affect the interindividual variation of warfarin dose,which can be sorteded into genetic factors and non-genetic factors.Non-genetic factors include age, gender,body surface area,drug interactions,dietary and morbid state.However, because the influence of non-genetic factors is limited,they are not the main reason of the variation of warfarin dose.Recently,with the rapid development of pharmacogenomics and elucidation of molecular mechanism of warfarin pharmacologic action,the effect of genetic factors on the variation of warfarin dose is more and more important.At present,there are more than 30 genes related to the pharmacodynamics and pharmacokinetics of warfarin,and CYP2C9 and VKORC1 are the two most important genetic factors.CYP2C9 is the major metabolic enzyme of warfarin.Up to now,there are 30 alleles of CYP2C9 be found,the common alleles are ~*1(wild type),~*2(Arg144Cys),~*3 (Ile359Leu).Patients carrying variant alleles have obviously lower enzymatic active and 2～3 times of bleeding risk than those carrying wild type alleles.VKORC is the target protein of warfarin.Warfarin exert its anticoagulation by inhibit the activity of VKORC,which function is reduces vitamin K 2,3-epoxide to the biologically active vitamin K hydroquinone.Several SNPs in VKORC1 gene were found to be related to the variation of warfarin dose,the most common SNPs were 1173C＞T in intron 1 and 3730G＞A in 3’ UTR.These SNPs were in strong Linkage Disequilibrium and the Haplotypes divided into two groups,A(H1 and H2) and B(H7,H8 and H9), associated with lower and higher warfarin dosage requirements respectively.The contributions of the two genes on the variation of warfarin dose were studied by many researchers,these studies showed CYP2C9 and VKORC1 could account 5-22%and 6-37%interindividual variation of warfarin dose respectively.In the recent years,several predictive warfarin stable dose algorithms have been founded based on genetic factors and non-genetic factors in same given populations.Before widely used in clinic,it is necessary for these algorithms to be validated for their safe and effectiveness by prospective random case-control study.At present,there are few prospective studies to validate these algorithms,which had insufficient persuasion for the defect of design or shortage of cases.Although there were warfarin stable dose algorithms for Chinese population,they are never be validated prospectively.The objectives of this study were to:1) Determine the relation of interindividual variation of warfarin dose and polymorphisms of VKORC1 and CYP2C9 in Chinese population;2)Develop a warfarin stable dose predictive algorithm incorporated genetic factors and non-genetic factors by a retrospective study;3) Validate the practicality and feasibility of this algorithm by a prospective random case-control study and search for a suitable personalized medicine model of warfarin in Chinese population;4) Develop a SYBR Green-based real-time PCR assay for SNP genotyping,which is simple,rapid,accurate,inexpensive and suitable for VKORC1 and CYP2C9 genotyping in clinical laboratory.Design and methodsThis study contains three sections:1) choose candidate genes and target SNPs, develope and evaluate genotyping method;2)retrospective study of warfarin personalized medicine and foundation of warfarin stable predictive algorithm;3) Validated the feasibility of this algorithm by a prospective study.We set up two mehtods,PCR/DHPLC and SYBR Green-Ⅰbased real-time PCR for genotyping VKORC1 T6484C and CYP2C9 A1075C.With direct DNA sequencing as gold standard,the sensitivity and specificity,time and cost of the two methods were compared,the one with more rapid,accurate and cost-effective was choosed for genotyping in this study.In the retrospective study,266 patients required warfarin stable dose were enrolled. We collected data on age,gender,height,weight,smoking,drinking,coadministration drugs and eating habit.A 5-ml blood sample of each patient was taken for genotype analysis.Correlations of stable warfarin dosage and each factor were analysised withα=0.05 as the size of a test.Factors that were not statistically significant were rejected.A multiple regression analysis was performed using factors with statistical significance to determin coefficient of determination(R~2),which means the proporation of the reason of interindividual warfarin dose could be explained by all the factors used.A warfarin stable dose predective algorithm was founded based on the multiple regression equation.In the prospective study,112 patients accorded with the selected standard and were first-time warfarin users were enrolled,divided into study group and control group randomly.In the study group,patients were genotyped for VKORC1 and CYP2C9 polymorphisms before taking warfarin,and the predicted doses were calculated by using warfarin stable dose predictive algorithm.The first three dosage of warfarin were taken according to the predicted doses,and the doses were adjusted based on the variation of INR values.In the control group,patients’ warfarin therapy were performed by tradictional model,and the warfarin doses were adjusted to stable dosage gradually.The frequency of INR monitoring was one time every from initial warfarin therapy to leave hospital,one time every week after discharge,and one time every month after stable dose acquired.Recored the time to stable dose and the time to adverise outcome,survival analyses were used to analysis experimental data.The endpoints were 1) dyas until an adverse outcome(defined as:any INR＞3.5,bleeding or venous thrombosis),and 2) days until stable dose acquired.The log-rank test compared the difference of time-to-event between study group and control group.A Cox proportional hazards-regression model generated HR(hazard ration) for study group and control group on the time to stable dose,and analysised the influence of each factors on endpoints.Results and discussionBoth the two SNP genotyping methods for VKORC1 and CYP2C9,DHPLC and real-time PCR assays were very accurate.The results showed complent concordance between DNA sequencing results and real-time PCR genotyping,and 99% concordance between DHPLC assay and DNA sequencing results.Compared with DHPLC,real-time PCR assay is convenient,time savings and inexpensive with relative throughput.It is suitable for VKORC1 and CYP2C9 genotyping in prospective application of warfarin therapyAmong the 266 patients in retrospective study,warfarin stable dosage ranged from 0.625 to 8.125mg/d.The mean of stable dose was 2.95±1.18mg/d and the coefficient of variance(CV) was 40%.The maximum dose exceeded the minimun dose by 13 times.Such further confirmed the existance of huge difference of warfarin among population.In the non-genetic factors,we selected eight factors,including gender,age, BSA,smoking,drinking,diabetes,indications and mean INR,to analysis their influence on interindividual warfarin stable dosage.The correlation analysis for the 8 factors and warfarin stable dosage(natural logarithmic transformation) shown gender, smoking,drinking,diabetes and indication had no relation to warfarin stable dosage, while age,BSA and mean INR had significant linear correlation to warfarin stable dosage.We found that the dose requirements fell with age,decreasing by approximately 0.2 mg per decade between the ages of 10 to 90 years irrsepective of other factors,and age accounted for 7.1%of interindividual variation of warfarin dosage.Warfarin dose requirement increased by 1.54mg per square meter of BSA irrsepective of other factors,and BSA accounted for 4.9%of interindividual variation of warfarin dosage.Correlation analysis showed that VKORC1 and CYP2C9 polymorphism had significant linear correlation to warfarin stable dosage.Irrsepective of other factors,VKORC1 and CYP2C9 accounted for 26.5%and 21%of interindividual variation of warfarin dosage respectively.Stepwise regression analysis was used to re-analysis the factors significantly related warfarin stable dose.The multiple regression shown age,BSA,VKORC1 and CYP2C9 were significant related to interindividual variation of warfarin dosage.The multiple regression equation of the optimized regression model was regarded as warfarin stable dose predictive algorithm,and this model accounted for 54.1%of interindividual variation of warfarin dosage.A total of 112 patients enrolled in prospective study.Of the total 95 patients completed the whole follow-up process,49 were in study group,and 46 were in control group.The mean time to stable dose of study group and control group were 27.4±1.8 days and 34.8±1.9 days respectively,and the median time were 24.0±1.6 days and 33.0±4.4 days respectively.The times to stable dose of the two groups had significant difference(P=0.011),which indicated that the time of stable warfarin dose adjustment could be shorten efficiently when patients taken warfarin according to the dose calculated by the predictive algorithm.Cox proportional hazards-regression analysis for group,gender,age,BSA,VKORC1 and CYP2C9 genotypes showen that only two factors,group and age had significant influence on time of stable warfarin dose adjustment(P=0.019,P=0.037).In the follow-up phase,the number of patients acquired stable dose in study group was 1.764(HR:1.764,95%CI:1.084-2.869, P=0.022) times than that in control group on the condition that the two groups had same total number of patients.There was a tendency that the time to stable dose prolonged with increasing age.In the four age groups,the time to stable dose of patients in the group of≥50 years was significantly increased than those of the other three groups.Of the 4 secondary indexes of prospective study,only the proporations of patients acquired stable dose had significant difference between study group and control group,the proporations were 81.6%and 63%respectively.The other three indexes,time of adjustment in hospital,incidence of adverse outcome and time to adverse outcome had no significant difference between the two groups.A total of 69 patients in the two groups acquired stable dose during follow-up phase.The mean predicted dose and mean actual dose of the 69 patients were 2.89±0.66 mg/d and 2.82±1.09mg/d respectively,with a mean overestimation of 0.07±0.81mg/d. predicted dose and actual dose had a significant linear correlation(r=0.676,P＜0.001). 66.7%patients’ predicted doses were within acceptable range.The results of prospective study indicated stable warfarin dose predictive algorithm incorporated genetic factors and non-genetic factors had high predictive effectiveness.ConclusionThe results of retrospective study indicated that age,BSA,VKORC1 and CYP2C9 polymorphisms had significant influence on interindividual variation of warfarin dose, and genetic factors play a key role in this variation.Stable warfarin dose predictive algorithm incorporated genetic factors and non-genetic factors accounted for 54.1%of the interindividual variation of warfarin dose.The results of prospective randomized case-control study indicated that the time of stable warfarin dose adjustment of the patients taken warfarin based on pharmacogenetics was apparently decreased than that of the patients taken warfarin traditionally.Predicted dose and actual dose of patients had a significant linear correlation.Our study confirmed that the warfarin dose predictive algorithm had a high clinical value.It is useful to improve the safety and effectivity of warfarin therapy.更多还原