Martinez Matias F, Alveal Enzo, Soto Tomas G, Bustamante Eva I, Ávila Fernanda, Bangdiwala Shrikant I, Flores Ivonne, Monterrosa Claudia, Morales Ricardo, Varela Nelson M, Fohner Alison E, Quiñones Luis A
Laboratory of Chemical Carcinogenesis and Pharmacogenetics (CQF), Department of Basic and Clinical Oncology (DOBC), Faculty of Medicine, University of Chile, Santiago, Chile.
Departamento de Ciencias y Tecnología Farmacéuticas, Facultad de Ciencias Químicas y Farmacéuticas, Universidad de Chile, Santiago de Chile, Chile.
Front Pharmacol. 2021 Mar 10;12:602676. doi: 10.3389/fphar.2021.602676. eCollection 2021.
Infections in hematological cancer patients are common and usually life-threatening; avoiding them could decrease morbidity, mortality, and cost. Genes associated with antineoplastics' pharmacokinetics or with the immune/inflammatory response could explain variability in infection occurrence. To build a pharmacogenetic-based algorithm to predict the incidence of infections in patients undergoing cytotoxic chemotherapy. Prospective cohort study in adult patients receiving cytotoxic chemotherapy to treat leukemia, lymphoma, or myeloma in two hospitals in Santiago, Chile. We constructed the predictive model using logistic regression. We assessed thirteen genetic polymorphisms (including nine pharmacokinetic-related genes and four inflammatory response-related genes) and sociodemographic/clinical variables to be incorporated into the model. The model's calibration and discrimination were used to compare models; they were assessed by the Hosmer-Lemeshow goodness-of-fit test and area under the ROC curve, respectively, in association with Pseudo-R. We analyzed 203 chemotherapy cycles in 50 patients (47.8 ± 16.1 years; 56% women), including 13 (26%) with acute lymphoblastic and 12 (24%) with myeloblastic leukemia. Pharmacokinetics-related polymorphisms incorporated into the model were rs2242480C>T and rs11231809T>A. Immune/inflammatory response-related polymorphisms were rs4696480T>A and rs1800796C>G. Clinical/demographic variables incorporated into the model were chemotherapy type and cycle, diagnosis, days in neutropenia, age, and sex. The Pseudo-R was 0.56, the -value of the Hosmer-Lemeshow test was 0.98, showing good goodness-of-fit, and the area under the ROC curve was 0.93, showing good diagnostic accuracy. Genetics can help to predict infections in patients undergoing chemotherapy. This algorithm should be validated and could be used to save lives, decrease economic costs, and optimize limited health resources.
血液系统癌症患者的感染很常见，通常会危及生命；避免感染可降低发病率、死亡率和成本。与抗肿瘤药物药代动力学或免疫/炎症反应相关的基因可以解释感染发生率的差异。构建基于药物遗传学的算法以预测接受细胞毒性化疗患者的感染发生率。在智利圣地亚哥的两家医院对接受细胞毒性化疗治疗白血病、淋巴瘤或骨髓瘤的成年患者进行前瞻性队列研究。我们使用逻辑回归构建预测模型。我们评估了13种基因多态性（包括9个与药代动力学相关的基因和4个与炎症反应相关的基因）以及纳入模型的社会人口统计学/临床变量。模型的校准和区分用于比较模型；分别通过Hosmer-Lemeshow拟合优度检验和ROC曲线下面积与Pseudo-R相关联来评估它们。我们分析了50名患者（47.8±16.1岁；56%为女性）的203个化疗周期，其中包括13名（26%）急性淋巴细胞白血病患者和12名（24%）髓细胞白血病患者。纳入模型的与药代动力学相关的多态性为rs2242480C>T和rs11231809T>A。与免疫/炎症反应相关的多态性为rs4696480T>A和rs1800796C>G。纳入模型的临床/人口统计学变量为化疗类型和周期、诊断、中性粒细胞减少天数、年龄和性别。Pseudo-R为0.56，Hosmer-Lemeshow检验的P值为0.98，显示出良好的拟合优度，ROC曲线下面积为0.93，显示出良好的诊断准确性。遗传学有助于预测接受化疗患者的感染情况。该算法应进行验证，并可用于挽救生命、降低经济成本和优化有限的卫生资源。
