This study showed that the distribution of SLCNs in different size groups was similar. It was generally believed that the lung cancers were mainly distributed in the upper lobes [29-31]. Concerning nodule location, most nodules, particularly the smaller ones, were not connected with the adjacent pleura. It appeared that the space between the tumor and pleura will subsist until the nodules grow large enough to involve it . It is possible that a SLCN is typically derived from the distal bronchus, and a distance exists between the lesion and pleura. These findings suggested that smaller nodules clinging to the pleura, which are usually detected on chest CT scan, were less likely to indicate lung cancer.
In this study, the smaller lesions (≤ 1cm) usually had different CT features with the bigger one (> 1cm). As nodule size increased, they acquired a more regular shape, and the margins and surrounding features also gradually increased. It may be because a nodule becomes more regular with an increase in size, and the limitation of the surrounding structure becomes more obvious. Lobulation, spiculation, pleural retraction, and vascular convergence are considered common signs of malignancy in lung cancer [6, 20-25]. However, for smaller nodules, traction and invasion of surrounding blood vessels and tissues, as well as tumor and peritumoral fibrosis were not obvious. As nodules increased, the invasiveness of the surrounding tissue also increased, and more signs were evident.
Siegelman et al.  reported that the incidence of coarse tumor–lung interface in lung cancer was significantly higher than that in benign lesions. In this study, tumor–lung interface in each group was mainly coarse. Additionally, the incidence of coarse interface increased with the increasing size of the nodule. This may be because tumor cells locally infiltrate the peripheral tissue, particularly the bigger nodules. However, the smaller the nodules, the higher the incidence of unclear interface, which may be related to the relatively sparse tumor cells in the peripheral areas of nodules. Therefore, for smaller solid nodules with unclear interface, the possibility of lung cancer cannot be completely excluded; follow-up is recommended to avoid an erroneous diagnosis.
The growth of solid lung cancer is a gradual process. The tumor cells gradually accumulate, and the lesion size continuously increases. Theoretically, tumor density will be more homogeneous on plain CT scan with an increase in the lesion size. In the present study, smaller nodules had a higher incidence of heterogeneous density, particularly those less than 1 cm in diameter. However, incidence of internal calcification, vacuole sign, or cavity was low in each group, like the previous reports [34, 35]. Therefore, small solid nodules with heterogeneous density can be selectively observed, and lung cancer should be highly suspected once their density increases and becomes homogeneous.
Beam-shaped opacity is a banded ground glass opacity, which is located at the side of the tumor close to the pleura in different directions. It is common in adenocarcinoma and highly significant in the diagnosis of lung cancer . This sign may be related to the traction of the surrounding lung tissue. In this study, the incidence of beam-shaped opacity was higher in the group D, but it was significantly lower in the group A. This implicates that small nodules are usually not able to cause significant changes in surrounding structures.
Halo sign was a nonspecific sign around solid pulmonary nodules, and its border was usually clear for lung cancer . In this study, a well-defined halo sign was mainly located on one side of the nodule, and its incidence was slightly higher in smaller lesions. Therefore, the smaller solid nodules without other features but halo sign should be suspicious for lung cancer.
Overall, regarding the smaller pulmonary nodules, follow-up looks like a good way to discriminate their nature based on CT feature changes. Recent study confirmed that quantitative image features (“radiomics”) could also discriminate benign from malignant pulmonary nodules . Additionally, quantitative radiomic signatures showed the potential to reveal and predict the tumor growth speed, and could identify the indolent from aggressive lung cancer . Thus, radiomics may provide a new way for evaluating and managing indeterminate pulmonary nodules in the future.
This study had several limitations. Evaluation of CT feature changing regularity for SLCNs was performed by comparing grouped nodules with different sizes rather than only following one group of lesions. Thus, results obtained here should be confirmed in clinical practice. Additionally, pathological types of SLCNs varied but showed no significant differences among different groups. Therefore, present results seem to represent a general, rather than a specific tumor type. It should be noted that some types within smaller samples may not conform to the general morphological development.