Abstract
BACKGROUND: Mediastinal lymph nodes (MLNs) can be enlarged due to various benign or malignant causes. Endoscopic ultrasound (EUS) is often employed for the acquisition of tissue specimens of the enlarged MLN (EMLN). AIM: To determine the causes, document the symptoms, and determine factors predicting good yield of EUS-guided EMLN biopsy. METHODS: All patients having EMLN (> 10 mm) on thoracic imaging and referred for EUS-guided biopsy were included in this retrospective observational study. Adequacy of the tissue specimen was assessed by the endoscopist with macroscopic on-site evaluation (MOSE) and then sent to a histopathologist for final diagnosis. Analysis for factors predicting good biopsy yield was then performed. RESULTS: Of the total 243 patients with EMLN, 131 (53.9%) were males. The mean age was 47.6 (± 14.7) and range 14-86 years. Commonest causes of EMLN were tuberculosis 82 (33.7%), anthracosis 53 (21.8%), neoplastic disease 43 (17.7%) and sarcoidosis 14 (5.8%). Among the 43 patients with neoplastic diseases: 40 had metastatic disease (breast 9, lung 7, pancreatobiliary 6, esophagus 5, kidney 5, hepatocellular 2, neuroendocrine 2, gastrointestinal stromal 1, undetermined 3) and 3 had primary malignancies (lymphoproliferative disorder 2, Kaposi sarcoma 1). Chest related clinical symptoms were absent in more than half the patients with tuberculosis 51/82 (62.2%), anthracosis 31/53 (58.5%) and neoplastic disease 28/43 (65.1%). Factors associated with good biopsy yield were: Subcarinal location of MLN (P = 0.026), MLN size > 12 mm (P < 0.0001), use of fine-needle biopsy (vs fine-needle aspiration) (P = 0.049) and satisfactory MOSE (P < 0.0001) on univariate analysis; while MLN size > 12 mm (P = 0.005) and satisfactory MOSE (P < 0.0001) on multivariate analysis. CONCLUSION: Tuberculosis, anthracosis and metastatic disease were the commonest causes of EMLN. More than half the cases with EMLN had no chest-related symptoms. Large MLN size and satisfactory MOSE observation predicted a good biopsy yield.