Abstract
Background Non-malignant conditions, including infections (such as tuberculosis [TB]), can mimic malignancy with regards to their uptake of 18F-fluorodeoxyglucose ((18)F-FDG) tracer utilized for positron emission tomography-computed tomography (PET-CT) scan, as part of the diagnostic and staging workup of cancer patients. This poses a diagnostic challenge, for which tissue sampling is decisive. In this study, we aimed to determine the underlying etiologies of (18)F-FDG-avid mediastinal lymph nodes among cancer patients in a TB-endemic demographic using endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and the respective sensitivity and specificity of PET-CT and EBUS in diagnosing malignancy. Methodology In this retrospective cross-sectional study, we analyzed the data of all cancer patients with (18)F-FDG-avid mediastinal lymphadenopathy on diagnostic PET imaging, who later underwent EBUS-TBNA between July 2013 and December 2018 at our center. Logistic regression analysis was used to determine the relative risk of lymph node characteristics with malignant TBNA cytology, based on which a risk stratification model was formulated. Results A total of 178 patients were included in this study, comprising predominantly males (60.7%). The primary malignancy was lung cancer in 33 (18.5%) patients, while 145 (81.5%) had non-lung cancer. A total of 214 (18)F-FDG lymph nodes were sampled, out of which TBNA revealed malignant cytology in only 44 (20.6%). The final diagnosis was malignancy, TB, and sarcoidosis in 42 (23.6%), 16 (9%), and 12 (6.7%) patients, respectively. Among the remaining, 98 (55%) patients were determined to have only reactive lymphadenopathy, of which 24 (24.5%) had nodal anthracosis, while TBNA was inadequate for the diagnosis in 10 (5.6%) patients. An increased risk of malignancy was associated with the size of lymph node [odds ratio (OR): 1.58 (confidence interval (CI): 1.19, 2.11; p = 0.001], the standard uptake value (SUV) of the lymph node on PET-CT [OR: 1.30 (CI: 1.15, 1.45); p = 0.001], and with primary lung malignancy [OR: 4.44 (CI: 1.96, 10.06); p = 0.001]. At an SUV cut-off value of 6.0, PET-CT had the sensitivity, specificity, positive predictive value, and negative predictive value of 73%, 70%, 49.3%, and 91.8%, respectively, for diagnosing malignancy, while the same for EBUS was estimated to be 93.3%, 100%, 100%, and 97%, respectively. Conclusions In addition to TB, benign etiologies including nodal anthracosis and sarcoidosis predominate as causes of (18)F-FDG-avid mediastinal lymphadenopathy in cancer patients of a TB-endemic demographic. The predictable risk of malignancy on PET imaging increases with nodal size, SUV, and lung primary malignancy; however, EBUS clearly demonstrates a higher sensitivity.