整合药物代谢组与药物基因组多元标志物的华法林个体化给药新模式研究

Rational use of warfarin has always been the key to the prognosis of cardiac valve replacement. However, there are a narrow therapeutic window, great individual differences, and adverse reactions such as prone to bleeding or thrombosis, resulting in a difficulty in dosage adjustment. So far, only 40% to 60% of individual differences in warfarin dose can be explained by the genetic background, suggesting that it is very necessary to explore the influencing factor of individual variations in drug response on a new level. Our previous studies have demonstrated that endogenous small molecule-based "pharmacometabonomics" can successfully predict individual pharmacokinetics variations of atorvastatin and clopidogrel after dosing. These researches indicated that endogenous small molecules may be potential biomarkers to predict the individual pharmacodynamic variation of warfarin. In the present study, patients who underwent cardiac valve replacement and took warfarin for the first time were selected as the research objects. The metabolomic profiles in preoperative baseline conditions, postoperative stress conditions, and initial /steady-state treatment period were measured. Endogenous molecular markers that could predict the individual variation of warfarin were screened out by a mathematical model combined with physiological significance, and its mechanism was explored. On the basis of this, pharmacogenomics-associated biomarkers were integrated to establish a clinically validated multivariate model and predict the pharmacodynamic response of warfarin and the dosage needed to achieve the therapeutic goals. The present study is expected to provide a novel approach for clinical precise dosing.

华法林治疗窗狭窄、个体差异大、极易因给药剂量不当出现严重不良反应。研究表明，遗传背景仅能解释约40%~60%的华法林个体差异，而药物代谢组学研究显示，药物体内暴露水平不仅与遗传多态性有关，还与基础代谢水平密切相关。我们的前期研究工作发现，基础状态下体内代谢小分子可作为潜在生物标志物，预测人体阿托伐他汀、氯吡格雷的药代动力学个体差异。提示内源性代谢小分子有可能成为预测华法林个体剂量差异的潜在标志物。本课题拟以心脏瓣膜置换术后首次服用华法林的患者为研究对象，通过测定和比较患者术前基础状态、术后应激状态、初始/稳态治疗期的代谢组水平，并与药效指标进行关联，筛选与药效相关的内源性小分子标志物；通过代谢通路、网络药理学分析初步探讨机理；再结合药物基因组学研究结果，整合基因标志物、代谢标志物构建多变量数学模型，预测华法林药效和个体化给药剂量，为临床精准给药提供新模式。

抗凝药华法林在初始治疗期的药效个体差异很大，用药剂量难以掌握，极易出现抗凝不足或过量的不良反应。使用传统的INR值监控试-错或是基因导向的稳态剂量公式，难以准确预测患者的初始剂量。如何在患者术后一周的院内治疗期将其华法林剂量尽快调整到安全有效的范围，是一个亟需解决问题。.本研究以105例心脏瓣膜置换术后首次服用华法林的患者为研究对象，（1）首先采用焦磷酸测序法测定华法林药效差异相关的SNP位点的基因型，结合临床信息建立基因导向的华法林初始治疗期的药效（INRday7）预测的多元线性回归模型。该模型能够解释47.9%、预测16.7%的个体差异。基因型和生理因素的预测效能有限。（2）采用LC/GC-QTOF-MS双平台检测受试者基础状态血清代谢轮廓，在总计6800个代谢物信号中鉴定出711种。将其与INRday7进行PLS回归，以VIP>1的268个代谢物带入代谢通路网络分析，结合数学上的显著性和生理意义，初步筛选潜在生物标志物。（3）采用标准品以标曲法对血浆中潜在生物标志物浓度进行绝对定性定量。最终优化的PLS模型以5个临床信息、1个基因型标志物和4个代谢标志物预测华法林INRday7，该模型该模型能够解释74%、预测65%的个体差异，理想预测率86%。与基因导向的传统模型相比，基础代谢组是优秀的强预测因子，是传统模型的强大补充。（4）为了提高研究结果的临床实用性，我们基于机器学习建立算法，用以预测受试者对华法林药效敏感性的分型（预测准确度达90.5%）。自编译的IniWarD软件通过固定剂量序列，求解落入期望药效区间的概率值，计算出不同个体的初始日均剂量推荐范围。与医生的经验给药剂量及患者服药7日后的实际INRday7进行比较，我们建立的整合药物基因组与药物代谢组多元标志物的机器学习模型，对药效敏感型区分更准确，得出的给药剂量策略具有合理性，不仅能预警华法林极端敏感或极端不敏感的病人，还能为医生对不同患者的个体化给药提供准确可靠的参考。该研究在指导精准用药方面提供了新思路和新途径，具有很高的发展潜力。