中国全科医学2018，Vol.21Issue(1)：52-57,6.DOI:10.3969/j.issn.1007-9572.2018.01.012
基于Logistic回归模型的低增生性骨髓增生异常综合征和再生障碍性贫血的鉴别诊断研究
Differential Diagnosis of Hypoplastic Myelodysplastic Syndrome and Aplastic Anemia Based on a Logistic Regression Model
摘要
Abstract
Objective To establish a Logistic regression model and evaluate its value in the differential diagnosis of hypoplastic myelodysplastic syndrome (hypo-MDS) and aplastic anemia (AA).Methods The clinical and laboratory data of 111 patients with hypo-MDS and 181 patients with AA hospitalized in the Blood Diseases Hospital, Chinese Academy of Medical Sciences between 2010 and 2016 were retrospectively collected. The sample data are randomly divided into model building groups and model verification groups according to 8:2. A Logistic regression model was established and used for the differential diagnosis of hypo-MDS and AA using the area under the receiver operating characteristic (ROC) curve, sensitivity, and specificity.Results The peripheral white blood cell and platelet counts, percentage of reticulocytes, proportions of CD3-CD56+ natural killer cells and CD3+CD57+T-large granular lymphocytes (LGL) in bone marrow cells, percentages of neutrophilic stab granulocytes and segmented neutrophilic granulocytes in bone marrow smears, and proportions of peripheral primitive myeloid cells, immature granulocytes, mature monocytes, and immature erythrocytes were significantly higher in patients with hypo-MDS than in those with AA (P<0.05). The red blood cell count, proportion of CD19+B cells in bone marrow cells, percentage of lymphocytes in bone marrow smears, and proportion of peripheral mature lymphocytes were significantly lower in patients with hypo-MDS than in those with AA (P<0.05). Unconditional binary Logistic regression analysis showed that patients with higher red blood cell counts and higher mature lymphocyte counts had a higher risk of developing AA, and those with higher proportions of segmented neutrophilic granulocytes, primitive myeloid cells, and CD3+CD57+T-LGL had a high risk of developing hypo-MDS (P<0.05). The Logistic regression model established was Logit(P)=1.293-0.584X1-0.060X2+0.055X3+0.561X4+0.059X5. The area under the ROC curve, sensitivity, and specificity were 0.852, 84.6％, and 78.9％, respectively, in the model establishment group and 0.899, 90.0％ and 84.6％, respectively, in the model verification group.Conclusion The Logistic regression model established for the differential diagnosis of hypo-MDS and AA had a high accuracy for predicting hypo-MDS and AA, facilitating the differential diagnosis of hypo-MDS and AA.关键词
骨髓增生异常综合征/贫血,再生障碍性/Logistic模型/诊断,鉴别Key words
Myelodysplastic syndromes/Anemia/aplastic/Logistic models/Diagnosis/differential分类
医药卫生引用本文复制引用
汪可可,武建辉,张晓雅,高涵,王国立,周莹,袁欣,王倩,曹英志,宋宇..基于Logistic回归模型的低增生性骨髓增生异常综合征和再生障碍性贫血的鉴别诊断研究[J].中国全科医学,2018,21(1):52-57,6.基金项目
河北省自然科学基金资助项目(H2017209172) （H2017209172）
