Page 110 - 《中国药房》2025年5期

·循证药学·

乳腺癌化疗致骨髓抑制风险预测模型的系统评价
Δ

1 #
刘 阳 1， 2＊ ，李红健 ，吴建华 ，刘学涛 ，焦 敏 ，于鲁海 （1. 新疆维吾尔自治区人民医院药学部，乌鲁木齐
1
2
1
1
830001；2.石河子大学药学院，新疆 石河子 832000）
中图分类号 R979.1 文献标志码 A 文章编号 1001-0408（2025）05-0612-07
DOI 10.6039/j.issn.1001-0408.2025.05.19

摘 要 目的 系统评价乳腺癌化疗致骨髓抑制的风险预测模型，为临床医疗工作者选择或开发有效预测模型提供科学的参考依
据。方法 系统检索中国知网、维普网、万方数据库、PubMed、Web of Science、Cochrane Library、Embase、Scopus数据库中有关乳腺
癌化疗致骨髓抑制风险预测模型的研究，检索时限为建库至2024年5月7日。由2名研究者独立筛选文献，根据预测模型系统评
价的严格评估和数据清单提取数据，并采用预测模型研究的偏倚风险评价工具分析纳入研究的偏倚风险和适用性。结果 共纳入
7项研究，包含12个模型，其中11个模型报告了受试者工作特征曲线下面积，为0.600～0.908；2个模型报告了校准方法；纳入模型
常见的预测变量为年龄、化疗前中性粒细胞计数、化疗前淋巴细胞计数、化疗前白蛋白含量。7项研究整体偏倚风险高（主要原因
为研究设计缺陷、样本量不足、变量处理方式不当、未报告缺失数据、模型评估指标缺乏等）但适用性良好。结论 乳腺癌化疗致骨
髓抑制风险预测模型的预测性能有待进一步提升，且模型整体偏倚风险高。未来的研究应遵循模型开发与报告规范，并结合机器
学习算法开发出预测性能好、稳定性强、偏倚风险低的风险预测模型，为临床提供决策依据。
关键词 乳腺癌；化疗；骨髓抑制；风险预测模型；系统评价

Systematic review of risk predictive models for chemotherapy-induced myelosuppression in breast cancer
1
2
1
LIU Yang ，LI Hongjian ，WU Jianhua ，LIU Xuetao ，JIAO Min ，YU Luhai（1. Dept. of Pharmacy， Xinjiang
1， 2
1
1
Uygur Autonomous Region People’s Hospital， Urumqi 830001， China；2. School of Pharmacy， Shihezi
University， Xinjiang Shihezi 832000， China）
ABSTRACT OBJECTIVE To systematically evaluate risk prediction models for chemotherapy-induced myelosuppression in
breast cancer， and provide a scientific reference for clinical healthcare workers in selecting or developing effective predictive
models. METHODS A systematic search was conducted for studies on predictive models of the risk of chemotherapy-induced
myelosuppression in breast cancer across the CNKI， VIP， Wanfang， PubMed， Web of Science， Cochrane Library， Embase， and
Scopus databases， with a time frame of the establishment of the database to May 7， 2024. Literature was independently screened by
2 investigators， data were extracted according to critical appraisal and data extraction for systematic reviews of predictive model
studies， and the risk of bias evaluation tool for predictive model studies was used to analyze the risk of bias and applicability of the
included studies. RESULTS There were totally 7 studies， comprising 12 models. Among them， 11 models indicated an area under
the subject operating characteristic curve of 0.600-0.908； 2 models indicated calibration. The common predictor variables of the
included models were age， pre-chemotherapy neutrophil count， pre-chemotherapy lymphocyte count， and pre-chemotherapy
albumin. The overall risk of bias of the 7 studies was high， which was mainly attributed to the flaws in the study design，
insufficient sample sizes， inappropriate treatment of variables， non-reporting of missing data， and the lack of indicators for the
assessment of the models， but the applicability was good. CONCLUSIONS The predictive performance of risk predictive models
for chemotherapy-induced myelosuppression in breast cancer remains to be further enhanced， and the overall risk of model bias is
high. Future studies should follow the specifications of model development and reporting， then combine machine learning
algorithms to develop risk predictive models with good predictive performance， high stability， and low risk of bias， so as to
provide a decision-making basis for the clinic.
KEYWORDS breast cancer； chemotherapy； myelosuppression； risk predictive model； systematic review

Δ 基金项目“天山英才”医药卫生高层次人才培养计划项目（No.
TSYC202301A028） 乳腺癌发病日益趋向年轻化，其发病率及致死率均
＊ 第一作者 硕 士 研 究 生 。 研 究 方 向 ：临 床 药 学 。 E-mail：
呈现逐年攀升的态势，对女性的生命健康构成了严峻威
2363901559@qq.com
胁，已成为重大的社会性健康问题。调查显示，2022 年
# 通信作者 主任药师，硕士生导师，硕士。研究方向：医院药学、
药物分析。E-mail：1523264450@qq.com 全球新发乳腺癌病例数达到约 230.9 万例，发病率位列

· 612 · China Pharmacy 2025 Vol. 36 No. 5 中国药房 2025年第36卷第5期