Indeterminate findings on imaging-guided biopsy: Should additional intervention be pursued?

OBJECTIVE. Our purpose was to determine the outcome of patients in whom the results of imaging-guided biopsies were indeterminate and to examine factors that may affect outcome. MATERIALS AND METHODS. During a 25-month period, 619 consecutive imaging-guided biopsies (CT, n = 268 [43%]; sonography, n = 351 [57%]) were performed on abdominal lesions. Of these biopsies, findings from 454 biopsies (73%) were positive for malignancy, findings from 21 biopsies (3%) yielded a benign diagnosis, and findings from 15 biopsies (2%) were nondiagnostic. Findings from the remaining 129 biopsies (21%) were considered indeterminate because the pathologic diagnosis revealed benign tissue, inflammation, stromal elements, or atypical cells. The frequency of malignancy in lesions of patients in this indeterminate group was determined. RESULTS. Thirty-eight of the 129 patients were lost to follow-up. Of the remaining 91 patients, 55 (60%) had true-negative (benign) results, and 36 (40%) had false-negative (malignant) results. A history of malignancy had no significant effect on the true- or false-negative rate (p = .799). However, the guidance technique had a statistically significant effect: CT and sonographic guidance yielded 25 (49%) and 11 (28%) false-negative results, respectively (p = .037). The number of needle passes, needle type and gauge, and biopsy site did not have a statistically significant correlation with the false-negative rate. The presence of atypical cells in the cytologic aspirate was highly predictive, with a 71% false-negative rate (p = .008). CONCLUSION. We found a high incidence of malignancy (40%) in lesions deemed indeterminate at the time of biopsy. Thus, close follow-up or additional intervention should be pursued in such cases. Sonographic guidance resulted in a statistically significant decrease in the false-negative rate when compared with CT guidance; however, a history of malignancy, the biopsy site, the needle gauge and type, and the number of passes did not have a significant effect on the false-negative rate.