Construction of an imaging diagnostic model based on computed tomograph signs for peripheral small cell lung cancer

TL;DR

This study builds a CT-based model to help distinguish small cell lung cancer from non-small cell lung cancer in small tumors, improving diagnostic accuracy.

Contribution

A novel CT imaging diagnostic model for peripheral small cell lung cancer with high sensitivity and specificity is developed.

Findings

Male gender, smooth edges, and less spiculation are independent predictive factors for pSCLC.

The model achieved an AUC of 0.842, with 84.2% sensitivity and 78.1% specificity.

Lymph node enlargement and fewer air bronchogram signs are significant indicators of pSCLC.

Abstract

To construct an imaging diagnostic model for peripheral small cell lung cancer (pSCLC) with a diameter of ≤ 3cm to improve differential diagnostic efficiency.

As a retrospective study, patients with pathologically confirmed lung cancer with tumor diameter ≤ 3 cm who were treated at the Guang’anmen Hospital South Campus, China Academy of Chinese Medical Sciences from May 2018 to May 2024 were retrospectively selected. All patients underwent computer tomography (CT) imaging. Patients with pSCLC (n=38) were identified first and then matched them to patients with peripheral non-small cell lung cancer (pNSCLC) (n=114) during the same period in a 1:3 ratio. Predictive factors of pSCLC were identified by logistic regression analysis, and a predictive model was constructed.

Logistic regression analysis confirmed that male gender, smooth edges, less spiculation sign, less air bronchogram sign, and lymph node enlargement are independent predictive factors for pSCLC. A predictive model that combines the above five predictive factors has high diagnostic efficacy for pSCLC. The receiver operating characteristic (ROC) analysis results showed the area under the curve AUC of 0.842 (95% confidence interval (CI): 0.759~0.925), with a sensitivity of 84.2% and specificity of 78.1%.

Male sex, smooth edges, less spiculation and air bronchogram signs, and lymph node enlargement identified by the CT scan were shown as independent predictive factors for pSCLC. Combining the above features has a high diagnostic efficacy for pSCLC.

Figures