Study Population
From March 2016 to January 2020, patients with pathologically confirmed benign GGNs or resolved GGNs were collected, and cases with pathologically confirmed malignant GGNs during the same period were also collected. Then cases that met the following criteria were enrolled in this study. Inclusion criteria: (1) patients with pulmonary GGNs manifesting as PSNs; (2) patients’ clinical data were complete. Exclusion criteria: (1) CT images with the thickness more than 1 mm; (2) presence of artifacts on CT images affecting evaluation. Finally, 117 patients with 119 PSNs were included in this study. The patients’ selection procedure was shown in Fig. 1.
CT examinations
All patients underwent non-contrast chest CT with a 128-slice multi-detector CT scanner (SOMATOM Definition Flash system, Siemens Medical Systems) while holding their breath after inspiration, and were scanned from thoracic inlet to lung base. CT examinations were performed with the following parameters: tube voltage, 110 – 120 kV; tube current, 50 – 150 mAs; beam pitch, 1.0; detector collimation, 0.6 mm; rotation time, 0.5 s; reconstruction thickness, 1.0 mm or 0.625 mm; reconstruction interval, 1.0 mm or 0.625 mm, and reconstruction kernel, medium-sharp algorithm.
Analysis of CT Features
CT images of all patients were reviewed by two radiologists with more than ten-year experience of chest CT interpretation, who were blinded to clinical data and pathological results. Any divergences of the two radiologists during evaluation were resolved by consensus. CT images were analyzed with lung window setting (window level, -600 HU; width, 1600 HU) by using Picture Archiving and Communication System and multiplanar reconstruction.
The overall CT features of each PSN were evaluated: (a) lesion size (mean of the longest diameter and the perpendicular diameter on axial images), (b) lesion area (the largest area of entire PSN on axial images), (c) location, (d) lesion shape (round, oval, or irregular), (e) lesion border (well-defined or ill-defined), (f) lesion margin (smooth or coarse), (g) lobulation, (h) spiculation, (i) air bronchogram, (j) bubble lucency, (k) pleural indentation, (l) pulmonary vessel changes (distorted, dilated, or both). For GGO component, its density and uniformity (homogeneous or heterogeneous) were also evaluated. CT features that were analyzed for solid component included (a) area (the largest area of internal solid component on axial images), (b) solid ratio (largest area of solid component divided by largest lesion area), (c) density, (d) number (solitary or multiple), (e) shape (round, oval, or irregular), (f) border (well-defined or ill-defined), (g) margin of well-defined solid component (smooth or coarse), (h) distribution (concentrated or scattered), (i) location (central, eccentric). We did not record the size for multiple and irregular solid components because it would preclude reliable and accurate evaluation.
Clinical and pathological data
Patients’ clinical and laboratory data were recorded through Electronic Medical Record System. Clinical data, including patient age, sex, smoking history (never-smoker, ex-smoker, or current smoker), smoking amount and history of cancer were recorded. Laboratory findings, such as white blood cell (WBC) count, blood eosinophil count, and presence of blood eosinophilia, were also recorded. These laboratory examinations were performed before operation and within a week after CT examination.
All existing histopathologic slides were reviewed by two pathologists and histopathologic analysis was performed according to the 2015 World Health Organization classification of tumors of the lung, pleura, thymus, and heart [14].
Statistical Analysis
Continuous data are expressed as mean ± standard deviation, whereas categorical variables are presented as numbers and percentages. Continuous data were analyzed by using the analysis of Variance or Wilcoxon rank sum test, and categorical data were analyzed by Pearson χ2 test or Fisher’s exact test. A P value less than .05 was considered to indicate a statistically significant difference. All statistical analyses were performed by using SPSS 20.0 (SPSS, Chicago, Ill).
Binary logistic regression analysis was performed to identify variables that could be used in differentiating benign from malignant PSNs. Because of multi-collinearity in some clinical data and CT features, the least absolute shrinkage and selection operator (LASSO) was used to further select features. Characteristics with a P value of less than .05 at univariate analysis were used as the independent variables for LASSO analysis, and then the selected variables were put into binary logistic regression analysis. Receiver operating characteristic analyses were conducted for the variables with statistically significant differences on logistic regression analysis.
Ethics
The study protocol was reviewed and approved by the Ethics Committee of The First affiliated Hospital of Chongqing Medical University (IRB No: 2019-062). The recommendations of the Declaration of Helsinki for biomedical research involving human subjects were also followed.