Work

Development of a Risk Prediction Model for Bone Metastasis in Lung Adenocarcinoma With a T1 Primary Tumor Based on Intratumoral and Peritumoral Radiomics

Year: 2026

Type: article

Abstract: PURPOSE: Bone metastasis significantly affects the prognosis of lung adenocarcinoma (LUAD) patients. This study aims to construct and validate a risk prediction model for bone metastasis in LUAD with a T1 primary tumor based on intratumoral and peritumoral radiomics features. MATERIALS AND METHODS: A total of 392 patients pathologically diagnosed with LUAD and a T1 primary tumor from two medical centers were retrospectively included (training cohort: n = 217, internal validation cohort: n = 93, external validation cohort: n = 82). Univariate and multivariate analyses identified independent risk factors for the clinicoradiologic model. Radiomics features were extracted from the gross tumor volume (GTV) and peritumoral tumor volume (PTV) in the training cohort CT images to establish intratumoral and peritumoral radiomics models. The optimal radiomics model was combined with clinicoradiologic features to develop a nomogram. Model performance was assessed using the area under the curve (AUC), calibration curves, and decision curve analysis (DCA). RESULTS: Among 392 LUAD patients with a T1 primary tumor, 147 had bone metastasis. The clinicoradiologic model incorporated three predictors: lymph node enlargement, pleural effusion, and carcinoembryonic antigen (CEA) levels. The PTV (-3 to 3 mm) radiomics model showed high discrimination performance, with an AUC of 0.810 (95% CI: 0.712-0.908) in the external validation cohort. The nomogram model demonstrated the highest discrimination performance, with an AUC of 0.884 (95% CI: 0.715-0.946), and showed acceptable calibration. CONCLUSION: In this retrospective two-center cohort of patients with LUAD and a T1 primary tumor, the combined clinicoradiologic-radiomics nomogram showed potential for stratifying the risk of synchronous bone metastasis at baseline evaluation. Further validation in larger and more representative cohorts is warranted before broader clinical application.

Source: Cancer Medicine

DOI: 10.1002/cam4.71996

Authors: Ting Li, Hui Zheng, Yang Guo, Kemeng Zhang, Meiyan Liao

Institutions: Wuhan University, Zhongnan Hospital of Wuhan University, Central Hospital of Wuhan

Language: English

Topic: Radiomics and Machine Learning in Medical Imaging

Subfield: Radiology, Nuclear Medicine and Imaging

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $4,700