编者按：11月16~18日，由美国心脏协会(AHA)主办的2019科学年会在美国费城召开。中山大学附属第一医院心血管内科庄晓东博士在大会上汇报了《Data-driven Trajectories of On-treatment Systolic Blood Pressure and Cardiovascular Risk in SPRINT and ACCORD: Similar Pattern in Different Trials》这一研究，以长期降压治疗时血压变化轨迹为切入点，采用新颖的线条轨迹聚类分析方法，发现“血压从基线高水平持续强化降压至较低水平”这一有害的血压变化轨迹是主要不良心血管事件的独立危险因素；目前，有关长期的降压治疗血压的动态变化及临床意义的研究相对空白，该研究为长期血压控制的长程动态管理提供新视角。
中山大学附属第一医院 庄晓东博士
将治疗中的收缩压（SBP）描述为轨迹可提供更多关于治疗效果的动态信息。然而，具有长期随访的治疗中的SBP轨迹及其对预后的影响尚未得到系统研究。
研究假设
这项研究的目的是通过收集SBP干预试验(SPRINT)和控制糖尿病患者心血管风险(ACCORD)的患者水平数据来描述治疗中SBP的轨迹，并探讨这些轨迹与心血风险之间的关系。
研究方法
本研究对SPRINT和ACCORD试验的患者水平数据进行二次分析。研究者分析了整个随访期间的治疗SBP变化，以描述对高血压治疗的反应，并使用新开发的R包(用于纵向数据聚类的kmedoids)识别不同的SBP轨迹。然后，检验不同治疗是SBP轨迹和心血管结果之间的关系。
研究结果
在SPRINT和ACCORD试验中，根据基线SBP模式分为三种不同轨迹:A组(高基线水平-下降趋势)，B组(中等基线水平-稳定趋势)和C组(中基线水平-增长趋势)。A组的心血管病发生率最高，而一级复合心血管终点发生率有显著性差异。调整可能的协变量后，以B组为参照，SPRINT研究中A组和C组的危险比分别为1.28（95%CI：1.03~1.59），1.03（95%CI：0.83~1.27）；ACCORD中分别为1.27（95%CI：1.08~1.49），0.88（95%CI：0.75~1.04）。即使在调整了SBP的均值或标准差和变异系数之后，这种相关性仍存在。
研究结论
结合基线血压和降压SBP轨迹可独立预测SPRINT和ACCORD的心血管结局，但需更多的研究来确定最佳和最简单的方法评估临床实践中的治疗轨迹。
上下滑动查看英文摘要
Abstract 15797: Data-Driven Trajectories Modeling of On-Treatment Systolic Blood Pressure and Cardiovascular Risks in Sprint and Accord: Similar Pattern in Different Trials
Abstract
Introduction: Characterizing on-treatment SBP as trajectories provides more dynamic information on treatment effectiveness. However， on-treatment SBP trajectories with long-term follow-up and its impact on prognosis has not been systematically studied.
Hypothesis: We aim to characterize trajectories of on-treatment SBP and to investigate the association of these trajectories with risk of CVD by pooling patient-level data from the Systolic Blood Pressure Intervention Trial (SPRINT) and Action to Control Cardiovascular Risk in Diabetes (ACCORD).
Methods: The current study was a secondary analysis of patient-level data from SPRINT and ACCORD trials. We analyzed on treatment SBP response during the entire follow-up period to characterize the response to hypertension treatment and to identify distinct on-treatment SBP trajectories using the newly developed R package (anchored kmedoids for longitudinal data clustering， akmedoids). We then examined the association between different on-treatment SBP trajectories and cardiovascular outcomes.
Results: There trajectories of long-term on-treatment SBP are identified and labeled by baseline SBP value followed by increasing or decreasing on-treatment pattern: group A (high-decreasing)， group B (moderate-stable ) and group C (moderate-increasing) in SPRINT and ACCORD trials. Moreover， a significant difference of the primary composite CVD incidence are observed with group A the highest incident of CVD. After adjustment for other potential covariates， using group B as the reference， the hazard ratio was 1.28 [95% confidence interval (CI)， 1.03-1.59， 1.03 (95% CI 0.83-1.27) in SPRINT， 1.27 (95%CI 1.08-1.49) and 0.88 (95%CI 0.75-1.04) in ACCORD for the group A and C， respectively. The association persisted even after adjustment for the mean or standard deviation， and coefficient of variation of SBP.
Conclusions: Trajectory with high baseline SBP level and decreasing on-treatment SBP independently predicts cardiovascular outcomes in SPRINT and ACCORD. More research is needed to define the best and simplest way to evaluate and minimize on-treatment trajectories in clinical practice.