Objective To investigate the invasive signs of early lung adenocarcinoma by thin-slice CT and to provide the basis for selecting rational clinical treatment.

Methods Thin-slice CT findings of 101 patients (42 males and 59 females, aged 28–75 (56.5±9.4) years) with lung adenocarcinoma confirmed by surgical histopathology examination in Huizhou Central People's Hospital from January 2015 to October 2020 were retrospectively analyzed. The imaging characteristics of nodules including internal characteristics (solid component, vascular sign, bronchial sign, and vacuole sign), marginal characteristics (smooth edges, lobulation sign, spiculated sign, and pleural indentation sign), long diameter, CT value, relative CT value, volume, and CT value/volume ratio. According to pathological subtypes, 25 patients were classified into the pre-invasion group, and 76 were classified into the invasive group. Two groups were compared, and Mann-Whitney U test and independent sample t test were used to compare the measurement data. χ² test and Fisher's exact probability method were used for counting data comparison. Receiver operating characteristic curves were drawn and the optimal critical value and the area under the curve (AUC) were calculated.

Results The pre-invasion group (25 cases) had 20 cases (80.0%) of pure ground-glass nodules and 5 cases (20.0%) of mixed ground-glass nodules. The invasion group (76 cases) had 13 cases (17.1%) of pure ground-glass nodules, 26 cases (34.2%) of mixed ground-glass nodules, and 37 cases (48.7%) of solid nodules. The probability of the internal characteristics of nodules including solid components, vascular signs, bronchial signs, and vacuole signs in the invasion group were significantly higher than that in the pre-invasion group (82.9% (63/76) vs. 20.0% (5/25), 26.3% (20/76) vs. 4.0% (1/25), 43.4% (33/76) vs. 16.0% (4/25), 32.9% (25/76) vs. 8.0% (2/25); χ²=4.14–33.82; all P<0.05). The probability of smooth edges in the pre-invasion group was significantly higher than that in the invasion group (68.0% (17/25) vs. 10.5% (8/76), χ²=33.36, P<0.001). The probability of lobulation signs, spiculated signs, and pleural indentation signs in the invasion group were significantly higher than that in the pre-invasion group (73.7% (56/76) vs. 32.0% (8/25), 71.1% (54/76) vs. 0, 47.4% (36/76) vs. 0; χ²=14.08, Fisher's exact probability method; all P<0.001). Statistically significant differences existed in long diameter (7.50 (6.50, 8.25) mm vs. 13.00 (11.00, 16.00) mm), CT value (−537.00 (−612.00, −418.00) HU vs. −61.00 (−318.25, 21.50) HU), relative CT value (−289.00 (−412.00, −210.50) HU vs. −758.50 (−839.00, −534.25) HU), volume (0.18 (0.14, 0.26) cm³ vs. 0.86 (0.44, 2.16) cm³), CT value/volume ratio (−2685.00 (−3564.00, −1972.00) vs. −48.19 (−422.14, 12.80)) between pre-invasion group and invasion group (Z=−6.51 to −5.43; all P<0.001). Optimal cutoff values existed between the pre-invasive and invasive lesions. When the long diameter of nodules was ≥8.75 mm, the sensitivity and specificity were 87.5% and 84.0%, respectively. When the nodule volume was ≥0.31 cm³, the sensitivity and specificity were 82.9% and 88.0%, respectively. When the CT value was −464 HU, the sensitivity and specificity were 89.5% and 72.0%, respectively. When the CT value/volume ratio was −1681.7, the sensitivity and specificity were 93.4% and 88.0%, respectively. The AUC of long diameter, CT value, volume, and CT value/volume ratio of nodules were 0.902 (95%CI: 0.843–0.962), 0.889 (95%CI: 0.824–0.955), 0.863 (95%CI: 0.784–0.942), and 0.936 (95%CI: 0.886–0.985), respectively.

Conclusions CT findings of the invasion signs of early lung adenocarcinoma are related to the volume, length, density, internal characteristics, and marginal characteristics of the nodules. The relevant signs need to be carefully observed in multiple planes and for a comprehensive analysis. Once found that it has a tendency to infiltrate, it should be operated as soon as possible.