Search Results

Objective:To establish and validate the prediction model for bone marrow metastasis (BMM) in malignant tumors by screening out laboratory multiparameters.Methods:This case-control study collected 444 cases of malignant tumor patients who were hospitalized in the First Hospital of Jilin University from March 2018 to March 2024, including 243 cases for model establishment set and 201 cases for model validation set. The model establishment set was divided into BMM positive group (81 cases) and BMM negative group (162 cases), and the model validation set was divided into positive group (67 cases) and a negative group (134 cases). We collected patients′ clinical information such as gender, age, clinical diagnosis, and results of 47 laboratory tests including routine blood analysis, coagulation, liver function, tumor markers, potassium, sodium, chloride, and calcium ion tests, bone marrow morphology, and bone marrow biopsy. BMM was taken as the outcome event, differencial variables were analyzed using inter group comparisons, the correlation among parameters was analyzed using Pearson correlation analysis, the risk factors for BMM were analyzed using multivariate conditional logistic regression analysis, to establish logistic model, followed by efficiency evaluation on BMM predictive model using receiver operating characteristic (ROC) curves.Results:In the model establishment set, Pearson correlation analysis of 28 parameters that differed between the BMM positive and negative groups revealed that the correlation coefficients of 17 parameters, including mean platelet volume (MPV), hematocrit (HCT), hemoglobin (HGB), and prothrombin time (PT), were no more than 0.6 ( P<0.05). Further multivariate conditional logistic regression analysis demonstrated that MPV, HGB, HCT, PT, red cell distribution width (RDW), platelet count (PLT), alkaline phosphatase (ALP), chloride (Cl -), and mean erythrocyte hemoglobin concentration (MCHC) were the risk factors of BMM occurence in malignancy [MPV ( OR=9.929, 95% CI 2.688-71.335), HCT ( OR=8.232, 95% CI 6.223-9.841), HGB ( OR=4.300, 95% CI 1.947-16.577), PT ( OR=3.738, 95% CI 1.359-11.666), RDW ( OR=1.995, 95% CI 1.275-3.807), ALP ( OR=1.025, 95% CI 1.012-1.045), PLT ( OR=1.014, 95% CI 1.002-1.031), MCHC ( OR=0.724, 95% CI 0.523-0.880) and Cl -( OR=0.703, 95% CI 0.472-0.967)]. In the model establishment set, combiation of risk factors provided an AUC of 0.943 (95% CI 0.898-0.987, P<0.001), a sensitivity of 86.3%, and a specificity of 89.2% for BMM prediction. In the model validation set, the AUC was 0.924 (95% CI 0.854-0.960, P<0.001), with a sensitivity and specificity of 86.7% and 83.8%, respectively. Conclusion:This study built and validated a multiple-parameter model for BMM, which may facilitate the timely detection of BMM and provide reference for decision making of bone marrow aspiration.