Multivariate analysis of chest CT features for predicting visceral pleural invasion in lung adenocarcinoma appearing as part-solid nodules

Objective  To investigate the clinical predictive value of chest CT features in the occurrence of visceral pleural invasion (VPI) in lung adenocarcinoma appearing as part-solid nodules (PSN).

Methods A retrospective analysis was performed on the clinical and imaging data of 206 patients with PSN lung adenocarcinoma who underwent surgical resection in Jiangmen Central Hospital from May 2016 to July 2022 (81 males and 125 females, age 59.5±10.9 years). The patients were divided into VPI positive and negative groups according to histopathological examination results. Chest CT features of the patients were reviewed, including the nodule location, boundary, morphology, diameter (total diameter, consolidation diameter, consolidation-to-tumor ratio (CTR)), lobulation sign, burr sign, vacuolar sign, air bronchial sign, emphysema sign, nodule and pleura relationship (NPR), consolidation and pleura relationship, distance between nodule center and pleura. The patients were further divided into four subtypes according to NPR. Univariate analysis was used to examine the differences in clinical data and CT features between the VPI positive and negative groups. Multivariate Logistic regression analysis was applied to establish the prediction model, and the receiver operating characteristic (ROC) curve was employed to calculate the area under the curve (AUC). Comparisons between groups were made using two independent samples t-test, Mann-Whitney U rank-sum test, chi-square test, or Fisher exact probability method.

Results Among the 206 patients, 47 were in the VPI positive group and 159 were in the negative group. No significant differences in gender and age (χ²=0.267, P=0.606; t=1.284, P=0.201) and in the location, total diameter, boundary, morphology, lobulation sign, burr sign, vacuolar sign, air bronchial sign, and emphysema sign (χ²=0.003–3.530, Z=−0.577, Fisher exact probability method, all P>0.05) were found between the two groups. The consolidation diameter (9.00 (7.00, 12.00) mm vs. 8.00 (5.00, 11.00) mm) and CTR (0.56±0.16 vs. 0.49±0.14) in the VPI positive group were significantly higher than those in the negative group (Z=−2.079, P=0.038; t=−2.672, P=0.008). Among the 206 patients, the proportions of VPI in different NPR subtypes were 7.32% (6/82) of type Ⅰ, 10.34% (6/58) of type Ⅱ, 47.22% (17/36) of type Ⅲ, and 60.00% (18/30) of type Ⅳ. Statistically significant difference in NPR subtypes was observed between the VPI positive and negative groups (χ²=52.040, P<0.001). Multivariate Logistic regression analysis showed that CTR (OR=38.159, 95%CI: 2.487–585.467, P=0.009) and NPR subtype (OR=3.110, 95%CI: 2.148–4.502, P<0.001) were independent risk factors for predicting the VPI status of PSN lung adenocarcinoma. The AUC and accuracy of CTR, NPR subtype, and their combined prediction models were 0.610 and 77.1%, 0.794 and 79.1%, and 0.822 and 71.4%, respectively.

Conclusions Chest CT features were helpful in the preoperative assessment of VPI status in PSN lung adenocarcinoma. CTR and NPR subtype were independent risk factors for predicting VPI status.