Endobronchial Ultrasound Guided Transbronchial Needle Aspiration of Solitary Lung Nodules and the Triple Diagnosis Technique: Does Triple Diagnosis Improve Diagnostic Accuracy?

Abstract

Background: Endobronchial ultrasound guided Transbronchial Needle Aspiration (EBUS-TBNA) is the predominant method for investigation of centrally located solitary pulmonary nodules. The method is associated with good to excellent diagnostic sensitivity and specificity with the positive predictive value of the test reaching 100% and reported negative predictive values for FNA of pulmonary nodules ranging from 53% to 97%. The impact of correlating cytologic results with imaging and clinical findings for improvement of negative predictive value has been poorly studied.

Materials and methods: An electronic search of the cytology records of the University of Missouri was made for all EBUS-TBNA specimens with surgical pathology follow-up. The associated imaging and clinical impressions were obtained. The clinical impression/assessment was the assessment given in the clinical summary section of the clinical note. Results of each technique were designated as benign, indeterminate, or malignant. Diagnostic triplets were constructed, and calculation of sensitivity, specificity, positive predictive value and negative predictive value was performed for each modality as well as the triplets.

Results: Search revealed a total of 254 cases with definitive surgical follow up. Thirty-two of these were excluded because they were either cytologically non-diagnostic or fell into one of the intermediate categories. One hundred sixty-three and 84 of these cases had definitive imaging or clinical interpretive results respectively. Definitive diagnostic triplets were obtained in 58 cases (22.8%). Cytology was associated with a sensitivity of 66.7% and specificity of 81.1%. Its positive predictive value was 87.7% and its negative predictive value was 54.5%. Imaging was associated with a sensitivity of 90.7%, a specificity of 22.2%, a positive predictive value of 75.4% and a negative predictive value of 47.6% and these statistics were 96.7%, 33.3%, 78.4%, and 80.0% respectively for clinical analysis. The positive predictive value of a triplet analysis was 92.9% and the negative predictive value was 100%.

Conclusions: Diagnostic triplets were obtainable in only a minority of cases, but they resulted in significant improvement in positive and negative predictive values.