Am J Med Sci. 2020 Apr 25:S0002-9629(20)30151-8. doi: 10.1016/j.amjms.2020.04.020. Online ahead of print.
ABSTRACT
BACKGROUND: The etiology of pleural effusions often remained unknown notwithstanding surgical pleural biopsy and further clinical observation. A better understanding of clinical characteristics of patients with idiopathic pleural effusion (IPE) may improve the ability to differentiate between IPEs and cytology-negative malignant pleural effusions (MPEs) and facilitate the identification of patients requiring invasive investigation. However, little is known about the clinical factors that can help distinguish patients with IPE from those with cytology-negative MPE.
MATERIALS AND METHODS: Patients who were diagnosed with IPE or cytology-negative MPE between 2010 and 2017 were enrolled in this retrospective study. Clinical, laboratory and radiologic characteristics were compared between patients with IPE and cytology-negative MPE. Diagnostic performances of predictors for IPE were assessed using receiver operating characteristic curves.
RESULTS: Of 146 patients undergoing pleural biopsy owing to cytology-negative pleural effusion of uncertain cause, MPE was confirmed in 54 patients. IPE was ultimately diagnosed in 22 patients. Multivariate analysis demonstrated that a minimal amount of pleural effusion (odds ratio [OR] = 12.41, P = 0.039), presence of pleural nodularity (OR = 0.01, P < 0.001) and pleural fluid carcinoembryonic antigen levels less than 14 ng/mL (OR = 87.59, P = 0.002) were independent factors for distinguishing IPEs from cytology-negative MPEs. A combination of the absence of pleural nodularity with pleural fluid carcinoembryonic antigen levels less than 14 ng/mL yielded an area under the curve of 0.94 (sensitivity = 91% and specificity = 96%).
CONCLUSIONS: Using these readily available parameters to identify IPE in patients with cytology-negative exudative effusion of unknown cause can help guide decision-making when choosing to perform an invasive pleural biopsy or to take a conservative approach.
PMID:32423747 | DOI:10.1016/j.amjms.2020.04.020