In recent years, the incidence and mortality rate of lung malignant tumors have ranked first in the world. Studies have shown that the 5-year recurrence rate after surgery for early-stage lung cancer is only 10%, and early detection of lung cancer with surgical treatment is the key to treating lung cancer (12). Early-stage lung cancer has no specific clinical manifestations, and its initial manifestation may only be isolated nodules in the lungs. Isolated nodules can be classified into solid nodules and subsolid nodules based on the different solid components within the PN. Solid nodules refer to nodules with a density that can block the small bronchi and blood vessels inside, while subsolid nodules refer to nodules with unclear or indistinct boundaries, and the density within the nodules cannot block the small bronchi and blood vessels inside (13, 14).
Studies have shown that compared to non-small cell lung cancer with subsolid nodules on imaging, lung lesions with pure solid nodules have a higher rate of lymph node metastasis and malignancy, and a worse prognosis (15, 16). Therefore, in clinical practice, solid PN require early and rapid diagnosis to avoid delaying the patient's condition. However, the judgment of imaging features is related to the experience of radiologists and thoracic surgeons, so the interpretation of imaging results may vary among doctors of different levels, and the imaging features of solid PN with a diameter ≤ 1 cm may not be obvious (17, 18). Therefore, having a unified diagnostic criteria and the ability to accurately identify small PN is an important prerequisite for screening early-stage lung cancer. Therefore, in this study, in the analysis of imaging features, the United Imaging Intelligence was used to uniformly identify the size and related morphology of PN in patients (19). However, relying solely on CT imaging features cannot completely differentiate between benign and malignant PN. Previous studies have shown that combining imaging features with more sensitive molecular markers can increase the positive diagnostic rate of lung cancer (20–22).
Related studies have shown that some tumor markers in serum can also assist in the diagnosis of lung cancer. However, traditional tumor markers such as CEA, NSE, and CA199 have low specificity and sensitivity in the diagnosis of early-stage lung cancer (22–25). In recent years, the emergence of liquid biopsy techniques represented by CTCs has provided new ideas for the early diagnosis of lung cancer. CTCs refer to tumor cells that detach from primary or metastatic lesions and enter the blood. CTCs are extremely rare in peripheral blood, and specific CTC markers are key to improving the detection rate of CTCs. Folate receptor positivity, as a specific marker of CTCs, has high sensitivity and specificity and can be highly expressed in tumor cells. Folate receptors are expressed very low in fallopian tubes, renal tubules, alveolar wall cells, choroid plexus, and uterus, and are not expressed in blood cells. The expression of folate receptors in lung cancer cells exceeds 78%, and folate receptors can recognize active CTCs, unaffected by the transformation of epithelial cells into mesenchymal cells. In this study, the CellCollector (GILUPI CellCollector, GILUPI) in vivo detection method was used to detect peripheral CTCs before surgery. Immunofluorescence staining was performed for CTC evaluation and PD-L expression analysis. In addition, CTCs captured by the CellCollector sampling probe were isolated and subjected to whole genome amplification and quality assessment. Qualified CTCs underwent NGS to determine copy number variations (CNV). In 2010, the American Joint Committee on Cancer (AJCC) Cancer Staging Manual proposed that CTCs could become a new indicator to help pathologists stage cancer, and CTCs have begun to be used in clinical practice to better formulate treatment plans for lung cancer patients (26–28). Previously, some people attempted to combine CTC evaluation with established screening methods, focusing on improving its specificity by combining low-dose CT screening plans with subsequent CTC evaluation. CTC counting evaluation was performed on patients with confirmed pulmonary "ground-glass" nodules and healthy controls. Only some patients with nodules had CTCs in their blood. Subsequent molecular analysis revealed that these CTCs had a "malignant tendency" (29). However, their study did not establish a model to predict the pathological nature of the nodules, and there was a lack of unified diagnostic criteria. We retrospectively evaluated the diagnostic efficacy of liquid biopsy combined with artificial intelligence in 76 suspected lung cancer patients, and the results showed that this predictive model had better predictive performance.
This study is based on the construction of a predictive model for PN using imaging AI and CTCs, and it has been confirmed to have good accuracy, stability, and applicability through internal and external validation, providing reference for clinical decision-making.