A retrospective study of 84 patients with spinal metastasis from prostate cancer (SMPCa) was performed.
The aim of this study was to predict the survival of patients with SMPCa by establishing an effective prognostic nomogram model, associating with the affecting factors and compare its efficacy with the existing scoring models.
Summary of Background Data.
Prostate cancer (PCa) is the second most frequently malignant cancer causing death in men, and the spine is the most common site of bone metastatic burden. The aim of this study was to establish a prognostic nomogram for survival prediction of patients with SMPCa, explore associated factors, and compare the effectiveness of the new nomogram prediction model with the existing scoring systems.
Included in this study were 84 SMPCa patients who were admitted in our spinal tumor center between 2006 and 2018. Their clinical data were retrospectively analyzed by univariate and multivariate analyses to identify independent variables that enabled to predict prognosis. A nomogram, named Changzheng Nomogram for Survival Prediction (CNSP), was established on the basis of preoperative independent variables, and then subjected to bootstrap re-samples for internal validation. The predictive accuracy and discriminative ability were measured by concordance index (C-index). Receiver-operating characteristic (ROC) analysis with the corresponding area under the ROC was used to estimate the prediction efficacy of CNSP and compare it with the four existing prognostic models Tomita, Tokuhashi, Bauer, and Crnalic.
A total of seven independent variables including Gleason score (P = 0.001), hormone refractory (P < 0.001), visceral metastasis (P < 0.001), lymphocyte to monocyte ratio (P = 0.009), prostate-specific antigen (P = 0.018), fPSA/tPSA (P = 0.029), Karnofsky Performance Status (P = 0.039) were identified after accurate analysis, and then entered the nomogram with the C-index of 0.87 (95% confidence interval, 0.84–0.90). The calibration curves for probability of 12-, 24-, and 36-month overall survival (OS) showed good consistency between the predictive risk and the actual risk. Compared with the previous prognostic models, the CNSP model was significantly more effective than the four existing prognostic models in predicting OS of the SMPCa patients (p < 0.05).
The overall performance of the CNSP model was satisfactory and could be used to estimate the survival outcome of individual patients more precisely and thus help clinicians design more specific and individualized therapeutic regimens.
Level of Evidence: 4