节点文献

高血压患者对于降压药物收缩压反应的数字模型研究

Establishing an in Silico Model to Predict the Modification of Systolic Blood Pressure Reaction to Pharmacological Intervention

【作者】 吴瑛

【导师】 李一石; 顼志敏; 刘玉清;

【作者基本信息】 北京协和医学院 , 内科学, 2010, 博士

【摘要】 研究背景高血压治疗的目的是最大限度地降低心脑血管疾病的发生率和死亡率,防止脑卒中、心肌梗死、心力衰竭和肾脏疾病的发生和发展。降压治疗的益处主要来自血压降低。目前高血压病的治疗仍然采用群体策略。国内外尚未见可供预测患者个体对于降压药物疗效反应的其他数学模型研究报道。研究目的运用建立数学模型的研究方法,探讨降压药物在高血压患者中收缩压反应的影响因素,为最终实现个体化药物治疗提供依据。研究方法选取MRC35-64、MRC65-74、STEP及SYST-EU四个随机、对照口服药物治疗高血压病的大规模临床试验,对于其纳入的高血压病患者,治疗组接受口服利尿剂、β受体阻滞剂或钙通道阻滞剂单种类药物治疗,对照组服用安慰剂,基线与药后第一次随访两者之间收缩压下降差值作为疗效指标。分析纳入病例的年龄、性别、身高、体重、吸烟、基础收缩压、舒张压、血清总胆固醇水平、既往心肌梗死病史等人口学及临床指标,分析药物治疗效果的影响因素,应用SAS■及R两种软件建立数学模型,探讨上述因素与药物降低收缩压效应的相关性。研究结果四个临床试验共31,140例原发性高血压患者,年龄60±11岁,男性14636人(47%),余为女性。平均基础收缩压168.0±16.8mmHg,舒张压92.1±12.54mmHg。药后第一次随访时,利尿剂、β受体阻滞剂及钙通道阻滞剂治疗组收缩压平均下降25.9±12.4mmHg、24.8±7.54mmHg、15.5±6.5mmHg。两种统计软件结果一致。建立的数学模型显示,降压药是收缩压降低的直接影响因素。利尿剂治疗组:年龄、基础收缩压和舒张压是降压疗效的影响因素,且三种因素协同作用于利尿剂的直接治疗,影响血压进一步下降。β受体阻滞剂组:基础收缩压、舒张压、身高正协同药物治疗作用,年龄负协同于药物降压疗效,非吸烟患者治疗后血压下降较吸烟者明显。钙通道阻滞剂治疗组:患者基础收缩压与患者体重为正协同于收缩压下降的影响因素。讨论与结论研究纳入了利尿剂、β受体阻滞剂、钙通道阻滞剂3类一线降压药物,患者涉及美国、欧洲多个国家和地区,4个临床随机对照研究均采用的是个体数据为基础,相关数学模型研究结果属首次报道。将患者临床特征(变量)纳入模型分析,发现这些变量除了直接与血压反应相关之外,又叠加作用于药物的治疗从而影响着血压反应结果。运用数学模型研究的方法预测降压药物的血压反应是可行的。但是应用不同的降压药物类别,显示其降压效果协同影响因素不同。由于影响血压的因素比较复杂,遗传学信息等尚不具备可供建立数学模型的数据库,其他可能还有影响因素未纳入本研究,故本模型研究有一定局限性,需要进一步完善和调整。

【Abstract】 BackgroundTreatment with blood pressure lowering drugs decreases cardiovascular risk mostly through blood pressure reduction. Selecting the best drug for a given hypertensive patient is difficult due to the number of drug classes. Current prescription strategy for these drugs relies upon some choice by the physician. The response to these drugs in terms of blood pressure is different from one patient to another.ObjectiveWe used the individual patient data from clinical trials, pooled in the INDANA data set, to explore whether blood pressure reduction was related to the baseline individual characteristics, and quantify these potential associations.MethodsWe used the data from patients with essential hypertension recruited in four randomized placebo-controlled clinical trials [Medical Research Council trial in mild hypertension (MRC35-64), Medical Research Council trial in older adults (MRC65-74). Systolic Hypertension in the Elderly Program (SHEP) and Systolic Hypertension in Europe trial (SYST-EU)]. Thiazide diuretics,β-blocker, and calcium channel blocker, three of six major BP lowering drugs were analyzed. Patients were all with the same first dosage of the drug in each trial. Age, body weight, height, level of TC, SBP and DBP when initialed and at first visit of follow-up, pharmacological treatment, gender, status of smoking, history of myocardium infarction were factors taken into model. Data managed by software SAS. Statistical analyses were performed with SAS and R software. Multiple regression analyses were used to evaluate the relationship between SBP fall and characteristics of patients. Significance threshold to keep an interaction in the model was 0.10, significance level of p<0.05 was used for other analyses. SAS and R developed the same model.ResultsIn all 31 140 patients (mean age 60±11 years,47% men) blood pressure at enrollment averaged 168.0mmHg systolic and 92.1mmHg diastolic. Among individual trials the SBP fall at visit 1 ranged from 16 to 26 mmHg for active treated group and from 8 to 17 mmHg in controlled.Initial SBP is the only modifier of treatment effect on BP response in the 3 BP lowering drug classes. Age and initial DBP were factors significantly correlated with SBP fall for diuretic and P-blocker. Smokers would receive less SBP fall compare to non-smokers in P-blocker active treated group. There is converse effect of age between the diuretic and P-blocker; older people seem sensitive with diuretic, while young people are sensitive with P-blocker. As to calcium channel antagonist class, only old patients recruited (age>60 years), and age is not a modifier.ConclusionsWe identified the significant modifiers for blood pressure response to pharmacological treatment effect; they are different between drug classes. Other factors including genotype information influencing blood pressure reaction to drugs still were not included in this model. This model need to be further improved and validated.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络