Influencing factors and prediction of growth heterogeneity in solid nodule non-small cell lung cancer based on artificial intelligence: a prospective study.

Background: The identification of rapidly growing solid nodules (SNs) through various preliminary examinations and their prompt removal can significantly improve the prognosis of patients with solid nodular lung cancer. However, previous studies have mostly focused on determining the nature of the solid nodules, with limited research on their growth heterogeneity. This study aimed to identify multi-dimensional factors influencing rapid nodule growth based on clinical, imaging, pathological, and genetic characteristics and provide a predictive model for solid nodule lung cancer growth. Methods: We prospectively analyzed 250 pathologically confirmed non-small cell lung cancer (NSCLC) nodules. Patients underwent preoperative thin-layer computer tomography (CT) scans with a median preoperative follow-up time of 75.5 (37.0, 273.3) days. All SNs in this study were divided into rapid (volume doubling time, VDT ≤200 days) and the slow growth group (VDT >200 days). Clinical data, imaging findings, pathological characteristics, and genetic mutations were analyzed. The Deep Wise Artificial Intelligence workstation was used to assess radiological qualitative features. Univariate and multivariate logistic regression analyses were used to determine the independent risk factors. Results: According to the VDT, 66.4% of the SNs grew slowly. Smoking history, CT value, and deep lobulation sign were risk factors for the rapid growth of nodules {area under the curve: 0.704 [95% confidence interval (CI): 0.636-0.771], sensitivity: 65.5%, specificity: 70.5%}. Pathologically, in the following order, squamous cell carcinoma had the fastest growth rate (squamous cell carcinoma > large cell neuroendocrine carcinoma > adenosquamous carcinoma > pleomorphic carcinoma > adenocarcinomas). Pathological histology type and degree of differentiation were risk factors for rapid growth (P=0.009, 0.006). Among the 168 nodules that underwent genetic testing, 75.6% had genetic mutations. Mutations in the epidermal growth factor receptor (EGFR) gene were the most common (43.4%). Mutations in tumor protein 53 (TP53) and anaplastic lymphoma kinase (ALK) mutations were enriched in adenocarcinomas with high-grade components (P=0.005, 0.03). Mutations in EGFR exon 21 L858R/19del and Kirsten rat sarcoma viral oncogene homolog (KRAS) differed between mucinous and non-mucinous adenocarcinomas (P<0.05). However, there was no significant correlation between nodule growth rates and gene mutations. Conclusions: Based on preoperative clinical and imaging data, rapidly growing nodules could be identified for early resection. Smoking history, CT values, and deep lobulation were critical predictors of rapid SN growth. Squamous cell carcinomas and poorly differentiated tumors accelerated nodule growth. Gene mutations drove the differentiation of NSCLC cells but did not regulate their growth rate.