OBJECTIVES Neutropenia is associated with the risk of life-threatening infections, chemotherapy dose reductions and delays that may compromise outcomes. This analysis was conducted to develop a prediction model for chemotherapy-induced severe neutropenia in lung cancer. MATERIALS AND METHODS Individual patient data from existing cooperative group phase II/III trials of stages III/IV non-small cell lung cancer or extensive small-cell lung cancer were included. The data were split into training and testing sets. In order to enhance the prediction accuracy and the reliability of the prediction model, lasso method was used for both variable selection and regularization on the training set. The selected variables was fit to a logistic model to obtain regression coefficients. The performance of the final prediction model was evaluated by the area under the ROC curve in both training and testing sets. RESULTS The dataset was randomly separated into training [7606 (67 %) patients] and testing [3746 (33 %) patients] sets. The final model included: age (>65 years), gender (male), weight (kg), BMI, insurance status (yes/unknown), stage (IIIB/IV/ESSCLC), number of metastatic sites (1, 2 or ≥3), individual drugs (gemcitabine, taxanes), number of chemotherapy agents (2 or ≥3), planned use of growth factors, associated radiation therapy, previous therapy (chemotherapy, radiation, surgery), duration of planned treatment, pleural effusion (yes/unknown), performance status (1, ≥2) and presence of symptoms (yes/unknown). CONCLUSIONS We have developed a relatively simple model with routinely available pre-treatment variables, to predict for neutropenia. This model should be independently validated prospectively.

中文翻译：

---

预测化疗引起的严重中性粒细胞减少症的风险：晚期肺癌个体患者数据的汇总分析。

目的中性粒细胞减少症与威胁生命的感染，降低化疗剂量和延误病情的风险有关。进行该分析以建立化疗诱导的肺癌中性粒细胞减少的预测模型。材料和方法包括来自现有合作小组III / IV期非小细胞肺癌或广泛性小细胞肺癌的II / III期合作试验的个体患者数据。数据分为训练和测试集。为了提高预测精度和预测模型的可靠性，在训练集上使用套索法进行变量选择和正则化。选择的变量适合于逻辑模型以获得回归系数。通过训练和测试集中的ROC曲线下的面积来评估最终预测模型的性能。结果该数据集被随机分为训练组[7606（67％）名患者]和测试组[3746（33％）患者]。最终模型包括：年龄（> 65岁），性别（男性），体重（kg），BMI，保险状态（是/未知），阶段（IIIB / IV / ESSCLC），转移部位的数量（1、2或≥3），单个药物（吉西他滨，紫杉烷类），化疗药物的数量（2或≥3），生长因子的计划使用，相关的放射疗法，既往疗法（化学疗法，放射线，手术），计划的治疗时间，胸腔积液（是/未知），工作状态（1，≥2）和症状的存在（是/未知）。结论我们开发了一个相对简单的模型，其中包含常规可用的预处理变量，预测中性粒细胞减少症。该模型应单独进行前瞻性验证。