Diagnostic value and accuracy of imprint cytology evaluation during image-guided core needle biopsies: Review of our experience at a large academic center.

BACKGROUND: Several studies have emphasized the value of on-site evaluation of imprint cytology (IC) performed on core needle biopsies (CNB) of breast, prostate, and lung, in terms of adequacy. The aim of this study was to investigate the diagnostic value and accuracy of rapid on-site IC of CNB specimens performed for liver, lung, lymph node, bone, and soft tissue masses to evaluate whether on-site preliminary diagnosis is sufficiently accurate to allow earlier, more efficient planning of ancillary studies with decreased turnaround time. METHODS: This morphology-based, retrospective study was approved by our Institutional Review Board. A total of 252 consecutive CNBs with on-site IC on masses of liver, lung, lymph node, bone, and soft tissue were included in this study. IC was reviewed by two cytopathology fellows and two board-certified cytopathologists who gave a categorical diagnosis (malignant/benign/atypical) and exact diagnosis when possible. Preliminary diagnoses were compared with corresponding histological CNB diagnoses. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated. RESULTS: Of the 252 cases reviewed, 30 cases were classified as atypical by IC and evaluated separately. Of the remaining 222 cases, IC classified an average of 154 (70%) as malignant, 54 (24%) as benign, and 14 (6%) as nondiagnostic. The corresponding distribution of histological diagnoses was 151 (68%) malignant and 71 (32%) benign. Overall correlation of correct IC diagnoses was 80%, with a correlation of 91% in malignant cases and 59% in benign cases. Sensitivity, specificity, PPV, and NPV were 96, 74, 92, and 87%, respectively. Diagnostic accuracy was 91%. There was no statistically significant difference in the accuracy of categorical diagnoses between IC and final histologic diagnosis. Atypical cases by IC were more likely to be malignant in lung and liver lesions (71% and 58%, respectively), than in lymph node, bone, or soft tissue lesions (17%, 0%, 0%, respectively). An exact diagnosis on IC was rendered in 113 (51%) cases, with an accuracy of 73%. Sensitivity, specificity, PPV, and NPV were 94, 41, 70, and 83%, respectively. CONCLUSION: Rapid on-site IC is a useful and valuable tool for evaluating adequacy of CNB as well as providing accurate information on a categorical basis (malignant versus benign), with greater diagnostic accuracy in cases of malignancy than for benign lesions. IC provides high sensitivity with high PPV in lung, liver, lymph node, bone, and soft tissue lesions. Exact preliminary diagnoses have good diagnostic accuracy. Considering the fundamentally different diagnostic and therapeutic approach based on histologic tumor type (e.g., in lung malignancies), on-site preliminary diagnosis may allow appropriate triaging of tissue for early planning of ancillary studies with decreased turnaround times. In addition, early diagnosis may reduce anxiety in patients and expedite treatment planning.