A prospective study comparing touch imprint cytology, frozen section analysis, and rapid cytokeratin immunostain for intraoperative evaluation of axillary sentinel lymph nodes in breast cancer

BACKGROUND: The intraoperative evaluation of axillary sentinel lymph nodes (SLNs) allows the surgeon to complete axillary dissection in 1 setting at the time of the primary breast surgery. However, to the authors' knowledge, there is no consensus regarding the optimal method for intraoperative evaluation of SLNs in breast cancer. The authors of this report prospectively compared touch imprint (TI) cytology with frozen section (FS) analysis and rapid cytokeratin immunostaining (RCI) of SLNs for the intraoperative evaluation of disease and compared the results with final pathologic examination (FP). METHODS: Patients with invasive breast carcinoma who were diagnosed with lymph node-negative disease (based on preoperative clinical and sonographic evaluation with or without fine-needle aspiration of the indeterminate lymph nodes) and who subsequently were scheduled for lymphatic mapping were eligible to participate in this prospective protocol. TI and FS analysis were performed on all SLNs, and the lymph nodes were stained by the he- matoxylin and eosin (H&E) method. RCI was performed using the enhanced polymer 1-step cytokeratin method. The results of TI, FS, RCI, TI plus FS, and FS plus RCI were compared with the results from FP, including 1 H&E stain and cytokeratin immunostain of the third level. RESULTS: One hundred patients with invasive mammary carcinoma were accrued to the study. Eighty-five tumors were the ductal type, 8 tumors were lobular, 5 tumors were mixed ductal and lobular, 1 was an adenoid cystic tumor, and 1 tumor was metaplastic carcinoma. Seventy-two tumors were staged clinically as T1N0M0, 25 tumors were staged as T2N0M0, and 3 tumors were staged as T3N0M0. Metastatic carcinoma was detected in the SLNs by 1 or more methods, including TI, FS, RCI, and FP, in 20 tumors, which included 12 macrometastases and 8 micrometastases. TI detected 8 of 12 macrometastases (67%), FS detected 12 of 12 macrometastases (100%), RCI detected 12 of 12 macrometastases (100%), and FP detected 12 of 12 macrometastases (100%). TI detected 1 of 8 micrometastases (13%), FS detected 3 of 8 micrometastases (38%), RCI detected 4 of 8 micrometastases (50%), and FP detected 6 of 8 micrometastases (75%). The sensitivities of TI, FS, RCI, TI plus FS, and FS plus RCI (with FP as the gold standard) were 50%, 72%, 78%, and 83%, respectively, and the sensitivities of the same intraoperative methods were 45%, 75%, 80%, and 85%, respectively, with detection of metastatic disease by any method as the gold standard. The specificities of the different methods (with FP as the gold standard) were 100% for TI and 97.5% for FS, RCI, TI plus FS, and FS plus RCI. The specificity of each method was 100% when the detection of metastatic disease by any method was regarded as the gold standard. Although the difference in sensitivity between FS and TI was not statistically significant (P =.08), the difference between RCI and TI bordered on significance (P =.046); however, FS analysis plus RCI was significantly superior to TI (P =.03) and produced results comparable to those of FP. CONCLUSIONS: The sensitivities of FS, RCI, TI plus FS, and FS plus RCI were better than the sensitivity of TI cytology of axillary SLNs. However, only the combination of FS and RCI was statistically superior to TI and generated results comparable to those of FP in SLNs. RCI can be completed within the time constraints for intraoperative use and, in conjunction with FS, can be useful for generating results closer to those generated by FP. FS analysis plus RCI have a role in the intraoperative evaluation of SLNs.