Distinguishing Organizing Pneumonia from Malignancy During Cancer Surveillance: Predominant Diagnostic Value of...

Background: Distinguishing organizing pneumonia from recurrent malignancy in patients undergoing cancer surveillance remains a major clinical challenge. Although inflammatory biomarkers have emerged as potential diagnostic tools, their incremental value beyond imaging-based assessment remains uncertain. Methods: In this retrospective single-center study, we evaluated 170 patients with prior malignancy who underwent surgical assessment for suspicious lung-only pulmonary lesions between 2013 and 2023. Histopathology confirmed organizing pneumonia in 61 patients and malignancy in 109 patients. Preoperative clinical, radiologic, Positron emission tomography/computed tomography (PET/CT), and inflammatory biomarker variables were incorporated into predefined clinical, radiology-PET/CT, biomarker, and integrated models using Least absolute shrinkage and selection operator (LASSO)-penalized logistic regression with internal cross-validation. Results: The radiology/PET model demonstrated excellent discrimination (AUC 0.927), closely approximating the performance of the integrated model (AUC 0.935). The clinical and biomarker-only models showed moderate discrimination (AUC 0.764 and 0.778, respectively). Adding inflammatory biomarkers to imaging-derived features resulted in only a minimal improvement in diagnostic performance (ΔAUC = 0.009, 95% CI -0.019 to 0.037), indicating no statistically significant incremental benefit. Calibration, decision curve, and predictor stability analyses consistently identified radiologic and PET/CT-derived variables as the principal sources of predictive information. Conclusions: In patients with suspicious pulmonary lesions during cancer surveillance, diagnostic discrimination was driven predominantly by imaging-derived features, whereas conventional inflammatory biomarkers provided limited incremental value. These findings support an imaging-centered approach to preoperative risk stratification and multidisciplinary decision-making.