[Comparison of touch imprint cytology of core needle biopsy and section histopathology in breast cancer diagnosis].

Objective: To evaluate the sensitivity, specificity of touch imprint cytology (TIC), and to compare its conformity rate with histopathology, to observe the consistence of immunocytochemistry (ICC) with immunohistochemistry (IHC), and to assess the diagnostic value of TIC prior to neoadjuvant chemotherapy for breast cancer.

Methods: 289 cases of TIC and 287 cases with core needle biopsy (CNB) histopathology accumulated from October 2005 to October 2008 in our hospital were included in this study. One hundred ninety cases TIC results were compared with that of final histopathology. 64 cases were tested for ER, PR, HER-2 by immunocytochemistry.

Results: Twenty-four benign cases and 263 malignant cases were diagnosed. 4 specimens were unsatisfactory. False negative rate and unsatisfactory rate were 1.4%, both, and false positive rate was 0.35%. The accuracy rate of TIC and CNB was 95.8% and 95.3%, respectively (P = 0.804). The sensitivity of TIC and CNB was 96.2% and 95.0% (P = 0.601), specificity 87.5% and 100% (P = 0.471) were found, when compared with the results of routine histopathology. 52 cases had a control with IHC of CNB in 64 ICC, and 43 cases had a final histopathology IHC. The ICC conformity rate of ER, PR, HER-2 with IHC of CNB was 86.5%, 75.0%, 78.8%, and that with IHC of final histopathology was 88.4%, 74.4%, 75.6%, respectively. The conformity rate of IHC between CNB and final histopathology was 83.7%, 74.4%, 76.5%, respectively. There was no significant statistical difference between them.

Conclusion: Compared with routine CNB histopathology, TIC has a high accuracy and sensitivity, and can provide a rapid and reliable cytological diagnosis to complement CNB for breast lesions. The conformity rates are high in ER, PR, HER-2 expression between ICC and IHC. ICC of TIC can be used to determine the estrogen and progesterone receptor levels in breast cancer before neoadjuvant chemotherapy.