A nomogram for differentiating false-negative from true-negative diagnoses following non-diagnostic CT-guided lung biopsy

This study aimed to develop and validate a nomogram that integrates clinical data, radiological findings, and biopsy procedural characteristics to distinguish false-negative from true-negative diagnoses in patients with non-diagnostic results following CT-guided lung biopsy. This retrospective study enrolled patients who underwent CT-guided lung biopsy and received non-diagnostic pathology results between January 2020 and December 2022. Using univariate and multivariate logistic regression, we developed two predictive models for false-negative results: a pathology-only model and a mixed model incorporating clinical and radiological data. Both models were visualized as nomograms. We enrolled 195 consecutive patients, randomly assigning 156 (80%) to the training cohort and the remaining 39 (20%) to the validation cohort. The mixed model demonstrated promising discriminative performance, with an area under the curve (AUC) of 0.93 (95% CI: 0.89–0.97) in the training set and 0.82 (95% CI: 0.65–0.99) in the small validation cohort of only 39 patients, which limited statistical power. Although the mixed model significantly outperformed the pathology-only model in the training set ( P < 0.001), the difference did not reach statistical significance in the validation set ( \(\:\text{P}=0.4\) ). This exploratory single-center mixed-model nomogram provides preliminary risk stratification for non-diagnostic CT-guided lung biopsies, large multicenter external validation is mandatory before routine clinical use. By enabling individualized risk stratification, this tool has the potential to guide clinical decision-making and support personalized management in this clinically challenging population.