A Personalized Medicine Approach to Improve the Prediction of Azathioprine Toxicity

摘要 硫唑嘌呤是一种免疫抑制药物，广泛用于治疗风湿性和其他炎症。 条件。然而，它的治疗指数很窄，而且临床上有显著副作用的频率很低。 与其使用相关的约50%。基于临床重要性和机制上的差异，这 该项目侧重于AZA最严重的两种不良反应：骨髓抑制和胰腺炎。目前， 临床医生仅限于硫代嘌呤甲基转移酶(TPMT)测试来预测患者服用硫唑嘌呤的风险 毒性。尽管有用，但TPMT基因多态只能解释四分之一的骨髓抑制病例 与硫唑嘌呤有关，但它们不能预测胰腺炎。最近的证据表明还有其他基因 变异体在硫唑嘌呤相关的副作用中起着重要作用。例如，NUDT15和人类白细胞抗原- DQA1*02：01-HLA-DRB1*07：01单倍型是骨髓抑制和胰腺炎的遗传决定因素。 分别进行了分析。然而，它们在常规临床实践中的有效性以及它们联合预测侧支循环的能力 AZA的效果尚不清楚。这一建议的首要假设是遗传风险得分 可以识别出出现硫唑嘌呤中毒的患者。使用最先进和新颖的技术和 资源，我们将进行遗传和基因表达关联分析，利用两个大型实践- 基于生物库：(1)Vanderbilt‘s BioVU，美国最大的基于实践的生物库之一，以及(2) 百万退伍军人计划(MVP)，目前正在登记、收集美国退伍军人的临床数据并对其进行基因分型。 在目标1中，我们将进行遗传关联分析，以发现骨髓抑制的新的遗传预测因子。 和服用硫唑嘌呤患者的胰腺炎。在目标2中，我们将检验这样一种假设，即新的遗传变异， 通过基因表达关联分析确定，可以预测AZA相关的骨髓抑制和胰腺炎。我们 将利用基因类型组织表达(GTEx)数据库预测基因表达。在《目标3》中，我们将 将AIMS 1和AIMS 2中识别的所有变异组合在一起以生成两个遗传风险分数(即，骨髓抑制风险 对于BioVU队列中的患者)。我们将进一步验证 独立的MVP队列。 该项目旨在通过构建两个遗传模型来推进精密医学倡议的目标 将预测硫唑嘌呤的严重和频繁的副作用。更好的预测能力将提供更好的治疗 为患者提供选择和先进的个性化药物，寻求提供“正确的药物，正确的剂量， 给合适的病人。“

Abstract Azathioprine is an immunosuppressive drug widely used for the treatment of rheumatic and other inflammatory conditions. However, it has a narrow therapeutic index, and the frequency of clinically significant side effects associated with its use is approximately 50%. Based on clinical importance and differences in mechanisms, this project focuses on two of the most serious adverse effects of AZA: myelosuppression and pancreatitis. Currently, clinicians are limited to thiopurine methyltransferase (TPMT) testing to predict patients' risk for azathioprine toxicity. Despite their usefulness, TPMT polymorphisms explain only one in four cases of myelosuppression associated with azathioprine, and they do not predict pancreatitis. Recent evidence suggests other genetic variants have important roles in azathioprine-related side effects. For example, NUDT15 and the HLA- DQA1*02:01–HLA-DRB1*07:01 haplotype are genetic determinants of myelosuppression and pancreatitis, respectively. Nevertheless, their usefulness in routine clinical practice and their combined ability to predict side effects of AZA remains unclear. The overarching hypothesis of this proposal is that genetic risk scores can identify patients who develop azathioprine toxicity. Using state of the art and novel techniques and resources, we will conduct genetic and gene expression association analyses, leveraging two large practice- based biobanks: (1) Vanderbilt's BioVU, one of the largest practice-based biobanks in the U.S., and (2) the Million Veteran Program (MVP), currently enrolling, collecting clinical data from, and genotyping U.S. Veterans. In Aim 1, we will conduct genetic association analyses to discover novel genetic predictors of myelosuppression and pancreatitis in patients taking azathioprine. In Aim 2, we will test the hypothesis that novel genetic variants, identified by gene expression association analyses, predict AZA-related myelosuppression and pancreatitis. We will predict gene expression by utilizing the Genotype Tissue-Expression (GTEx) database. In Aim 3, we will combine all variants identified from Aims 1 and 2 to generate two genetic risk scores (i.e., myelosuppression risk and pancreatitis risk) for patients in the BioVU cohort. We will further validate the genetic risk scores in the independent MVP cohort. This project aims to further the goals of the Precision Medicine Initiative by constructing two genetic models that will predict serious and frequent side effects of azathioprine. Better prediction capacity will offer better treatment options for patients and advance personalized medicine, which seeks to deliver “the right drug, at the right dose, to the right patient.”