Objectives: The objective was to identify predictors of true negatives in lung nodules (LNs) with computed tomography-guided percutaneous biopsy (CTPB)-based benign pathological results.
Materials and methods: We included 90 total patients between January 2013 and December 2017 that had CTPB-based nonspecific benign pathologies and used these patients as a training group to accurately identify true-negative predictors. A validation group of 50 patients from January 2018 to June 2019 to confirm predictor reliability.
Results: CTPB was conducted on 90 LNs from the training group. True-negative and false-negative CTPB-based pathologies were obtained for 79 and 11 LNs, respectively. CTPB-based benign results had a negative predictive value of 87.8% (79/90). Univariate and multivariate analyses revealed younger age (P = 0.019) and CTPB-based chronic inflammation with fibroplasia (P = 0.010) to be true-negative predictors. A predictive model was made by combining these two prognostic values as follows: score = -7.975 + 0.112 × age -2.883 × CTPB-based chronic inflammation with fibroplasia (0: no present; 1: present). The area under receiver operator characteristic (ROC) curve was 0.854 (P < 0.001). To maximize sensitivity and specificity, we selected a cutoff risk score of -0.1759. The application of this model to the validation group yielded an area under the ROC curve of 0.912 (P < 0.001).
Conclusions: Our predictive model showed good predictive ability for identifying true negatives among CTPB-based benign pathological results.
Keywords: Benign; biopsy; false; lung nodule; true.