急性淋巴细胞白血病患儿化疗相关性口腔黏膜炎的风险预测模型构建与验证

Construction of a risk prediction model for chemotherapy-induced oral mucositis in children with acute lymphoblastic leukemia and its validation

目的调查分析急性淋巴细胞白血病(ALL)患儿发生化疗相关性口腔黏膜炎(CTOM)的危险因素,构建风险预测模型并验证其预测效能。方法回顾性收集广东省2所三甲医院的470例ALL患儿的数据资料,将2020年1月-2022年4月一所医院收治的376例ALL患儿作为建模组,将2021年12月-2022年7月另一所医院收治的94例ALL患儿作为验证组。采用logistic回归分析收集的资料数据,确定ALL患儿发生CTOM的危险因素,构建风险预测模型,并进行内外部验证。结果ALL患儿CTOM发生率为45.1%,影响因素有:化疗周期≥5个疗程、此次化疗药物含甲氨蝶呤/柔红霉素/阿糖胞苷、携带单纯疱疹病毒1型、感染白色念珠菌、临床危险度为高危、年龄≥6岁、此次化疗前进行预防性漱口的频率≥3次(P<0.05)。建模组的C统计量为0.769,约登指数最大值为0.409,灵敏度为59.9%,特异度为81.0%,内外部验证的C统计量分别为0.772和0.865,校准曲线显示模型拟合良好,决策曲线分析表明模型具有一定的临床可用性。结论ALL患儿CTOM发生率较高,构建的风险预测模型具有较好的预测效能。

Objective To investigate and analyze the risk factors of chemotherapy-induced oral mucositis(CTOM)in children with acute lymphoblastic leukemia(ALL),construct a risk prediction model and validate its predictive performance.Methods The data of 470 children with ALL from two tertiary hospitals in Guangdong province were retrospectively collected.A total of 376 children with ALL admitted to one tertiary hospital from January 2020 to April 2022 were used as the modeling group,and 94 children with ALL admitted to another tertiary hospital from December 2021 to July 2022 were used as the validation group.Logistic regression was used to analyze the collected data to determine the risk factors for CTOM in children with ALL,construct a risk prediction model,and conduct internal and external verification.Results The incidence of CTOM in children with ALL was 45.1%,and the influencing factors were:chemotherapy cycle≥5 courses,chemotherapy drugs containing methotrexate/daunorubicin/cytarabine,carrying herpes simplex virus type 1,infection with Candida albicans,the clinical risk was high risk,age≥6 years old,and the frequency of preventive mouth washing before chemotherapy was≥3 times(P<0.05).The C-statistic value of the modeling group was 0.769,the maximum Youden index was 0.409,the sensitivity was 59.9%,and the specificity was 81.0%.The C-statistic values for internal and external validation were 0.772 and 0.865,respectively.The calibration curve showed that the model fit well,and the decision curve analysis showed that the model had certain clinical usability.Conclusion The incidence of CTOM in children with ALL is relatively high,and the constructed risk prediction model has good predictive performance.