Thin-needle aspiration biopsy: methods of increasing the accuracy of a negative prediction.

A total of 164 consecutive patients were studied prospectively in the hope of increasing the accuracy of a negative prediction of transthoracic thin-needle biopsy. The protocol included pre-biopsy imaging with conventional and computed tomography, repeated sampling if no malignant cells were found, and specialized staining and culture methods. Forty-four patients with no evidence of malignant cells were divided into "benign specific" [25] and "benign nonspecific" groups [19]. A variety of bacterial and fungal infections, benign tumors, and other benign specific diagnoses were made. The accuracy of a negative prediction for the benign specific group was found to be 1.00 (100%), while that for the benign nonspecific group varied from 84% to 95% depending on whether the 2 patients lost to follow-up had benign disease.