Establishment and validation of a predictive model for bone metastasis in prostate cancer patients based on multiple immune inflammatory parameters

Objective: This study aims to establish and validate a predictive model for bone metastasis in prostate cancer patients based on multiple immune inflammatory parameters. Methods: In this retrospective study, 162 prostate cancer patients who met the inclusion criteria were selected by Urology Surgery, Shaanxi Provincial People's Hospital. Based on the medical record number of patients and the random number table method, 40 patients were randomly included in a validation group, and the rest were in a modeling group. The patients in the modeling group were divided into a metastatic group (n=67) and a non-metastatic group (n=55) according to the whole-body bone imaging results. Results: The predictive model was established based on the results of Logistics regression analysis: Logit (P) = -5.341 + 0.930*total Gleason score + 1.426*total prostate specific antigen + 0.836*neutrophil-lymphocyte ratio + 0.896*platelet lymphocyte ratio + 0.641*lymphocyte/monocyte ratio + 0.750*albumin/globulin ratio. ROC analysis showed that the areas under the curve of the predictive model for bone metastasis in the modeling and validation groups were 0.896 and 0.870, respectively. Hosmer-Lemeshow test showed that P=0.253, indicating a high degree of the fitting. External verification results showed that the C-index for predicting prostate cancer bone metastasis in the predictive model established in this study was 0.760 (95% CI: 0.670-0.851). Conclusion: The bone metastasis predictive model based on the multiple immune inflammatory parameters (neutrophil-lymphocyte ratio, platelet lymphocyte ratio, lymphocyte/monocyte ratio and albumin/globulin ratio) in prostate cancer patients can reasonably predict the occurrence of bone metastasis and is well worth clinical application.