Malignancy risk for the categories: Non-diagnostic, benign, atypical, suspicious, and malignant used in the categorization of endobronchial ultrasound guided-fine needle aspirates of pulmonary nodules.

BACKGROUND: Endobronchial ultrasound-guided fine-needle aspiration (EBUS-FNA) is frequently used for the workup of pulmonary nodules. While no universally accepted diagnostic classification exists, many cytopathologists use the categories: Non-diagnostic, benign, atypical, suspicious and malignant. Sensitivity and specificity for the EBUS technique have been documented, but little information is available for malignancy risk associated with these categories.

METHODS: Departments of Pathology records at the University of Utah and University of North Carolina, Chapel Hill were searched for EBUS-FNAs of pulmonary nodules. Cases with surgical follow-up were selected. Cytologic diagnosis and subsequent surgical diagnosis were correlated and malignancy risk calculated for each category. Sensitivity and specificity were calculated.

RESULTS: 155 EBUS-FNAs with surgical follow-up were obtained. Risks of malignancy were: Non-diagnostic 40%, benign 24%, atypical 54%, suspicious for malignancy 82% and malignant 87%. Sensitivity and specificity were 81% and 84% respectively for surgically confirmed cytologic diagnoses when indeterminate categories were excluded.

CONCLUSION: The diagnostic categories are associated with increasing risk of malignancy running from non-diagnostic to malignant. The non-diagnostic category has a significant risk of malignancy. While the risk of malignancy for a benign diagnosis is substantial (24%), it is significantly less than that associated with an atypical or suspicious diagnosis. A suspicious diagnosis carries a risk for malignancy essentially the same as a malignant diagnosis. The categories atypical and suspicious appear to have substantially different risks for malignancy (54% vs.82%). The atypical category has twice the risk of malignancy as benign. This risk stratification may be useful for patient management.