A Predictive Model to Differentiate Between Second Primary Lung Cancers and Pulmonary Metastasis

Abstract

Rationale and objectives: To develop and validate a nomogram for differentiating second primary lung cancers (SPLCs) from pulmonary metastases (PMs).

Materials and methods: A total of 261 lesions from 253 eligible patients were included in this study. Among them, 195 lesions (87 SPLCs and 108 PMs) were used in the training cohort to establish the diagnostic model. Twenty-one clinical or imaging features were used to derive the model. Sixty-six lesions (32 SPLCs and 34 PMs) were included in the validation set.

Results: After analysis, age, lesion distribution, type of lesion, air bronchogram, contour, spiculation, and vessel convergence sign were considered to be significant variables for distinguishing SPLCs from PMs. Subsequently, these variables were selected to establish a nomogram. The model showed good distinction in the training set (area under the curve = 0.97) and the validation set (area under the curve = 0.92).

Conclusion: This study found that the nomogram calculated from clinical and radiological characteristics could accurately classify SPLCs and PMs.

Keywords: nomogram; predictive model; pulmonary metastasis; radiology; second primary lung cancer.