Comparison of the Brock model and LU-RADS in differentiating benign and malignant sub-solid pulmonary nodules

Abstract Background: The Lung Imaging Reporting and Data System (LU-RADS) is one of the most widely used tools assessing pulmonary nodules in clinical practice. The Brock model is a predictive model that shows good accuracy for determining likelihood of malignancy in pulmonary nodules, but both have not been validated and compared in sub-solid pulmonary nodules (SSNs); thus, the purpose of this study was to compare the Brock model and LU-RADS in differentiating malignant from benign SSN. Methods: The clinical data of patients with SSN who underwent surgical resection and were pathologically confirmed from January 2018 to December 2021 were retrospectively analyzed. According to the pathological results, the patients were divided into two groups, the benign SSN group and the malignant SSN group. Their clinical data were then statistically analyzed. The malignant probability of SSN was calculated by the Brock model. The LU-RADS category of SSN was determined by two radiologists independently. The receiver operating characteristic curves of the Brock model and LU-RADS were drawn, and the area under curve (AUC) was compared. Results: A total of 133 patients with SSN were included in this study. LU-RADS categories 4A and 4B of the malignant SSN group were more than the benign SSN group (56 versus 4, P < 0.05); the malignant probability of the malignant SSN group was significantly higher than that of the benign SSN group (0.21 versus 0.06, P < 0.05). The Brock model has a good correlation with LU-RADS ( r = 0.75, P < 0.01) and had a comparable performance in diagnosing lung cancer for the SSN (Brock versus LU-RADS, AUC: 0.83 versus 0.78, P = 0.16). Sub-group analysis showed that the Brock model had a higher performance in diagnosing malignancy for the mixed ground-glass nodules than LU-RADS (Brock versus LU-RADS, AUC: 0.92 versus 0.85, P = 0.03), but both had a similar lower performance in the detection of malignancy from the pure ground-glass nodules (Brock versus LU-RADS, AUC: 0.59 versus 0.55, P = 0.66). Conclusion: The Brock model was more suitable for distinguishing malignant from benign mixed ground-glass nodules. But both the Brock model and LU-RADS had poor performance in distinguishing malignant from benign pure ground-glass nodules.