Diagnostic value of bone metabolic markers for bone metastasis of non-small cell lung cancer analyzed by Logistic regression combined with ROC curve

Abstract

Objective To investigate the diagnostic value of bone metabolic markers (BGP, β-CTX and PINP) for bone metastasis of non-small cell lung cancer (NSCLC) analyzed by Logistic regression combined with ROC curve. Methods A total of 65 patients with stage Ⅳ NSCLC were enrolled in this study. The patients were divided into two groups based on radiological imaging, includirg bone metastasis group (30 cases) and non-bone metastasis group (35 cases). The serum concentrations of BGP, β-CTX and PINP were measured by electrochemical method. Logistic regression and ROC curve were applied to analyze the data and evaluate the diagnostic values. Results The concentrations of β-CTX [(0.54±0.39) ng/ml] and PINP [(103.64±81.86) ng/ml] were significantly higher in bone metastasis group than those [(0.31±0.16) ng/ml and (48.37±27.76) ng/ml, respectively] in non-bone metastasis group (P 0.05). The area under the ROC curve (AUC) for β-CTX and PINP was 0.662 and 0.678, respectively. The AUC for the new predictive variables created by Logistic regression was 0.761. Conclusion Combined detection of β-CTX and PINP in the serum and application of Logistic regression combined with ROC curve analysis can increase diagnostic accuracy on bone metastasis of NSCLC. Key words: Carcinoma, non-small-cell lung; Neoplasms metastasis; β-isomerized C-terminal telopeptides of type Ⅰ collagen; N-terminal propeptide of type Ⅰ procollage; Logistic models; ROC curve