Establishment and verification of a predictive model for bone metastasis in patients with non-small cell lung cancer based on peripheral blood CX3CL and CCL28.

Abstract

To establish a predictive model of bone metastasis in patients with non-small cell lung cancer (NSCLC) using peripheral blood CX3CL and CCL28, and to verify its application value. We retrospectively gathered clinical data from 210 patients with NSCLC treated at our institution between April 2021 and December 2023. These patients were stratified into two groups based on the presence of bone metastases: a bone metastasis group (n = 49) and a non-bone metastasis group (n = 161). A logistic regression model was developed to predict bone metastasis and to evaluate the model's predictive performance. Multivariate logistic regression analysis identified age (OR = 6.689, P < 0.001), carcinoembryonic antigen (CEA, OR = 5.699, P < 0.001), CX3CL1 (OR = 5.418, P < 0.001), and CCL28 (OR = 7.692, P < 0.001) as independent predictors of bone metastasis in NSCLC patients. The receiver operating characteristic (ROC) curve analysis yielded an area under the curve (AUC) of 0.794 for both the modeling and validation cohorts. Decision curve analysis (DCA) indicated a superior net benefit of the model. Calibration curves confirmed close concordance between predicted and observed probabilities of bone metastasis. The Hosmer-Lemeshow test yielded a chi-square statistic of 4.743 with a P-value of 0.178, suggesting a good fit. The predictive model utilizing serum levels of CX3CL1 and CCL28 demonstrates robust predictive accuracy and efficacy for bone metastasis in patients with NSCLC.