Context.—: Touch preparation (TP) alone is discouraged for intraoperative lymph node (LN) assessment in the neoadjuvant setting (NAS) owing to overall low sensitivity in detecting metastatic breast cancer.

Objective.—: To compare the sensitivity, specificity, and negative predictive value of intraoperative LN assessment via TP and examine potential causes of discrepancies along with the clinical, radiologic, and pathologic parameters in the neoadjuvant setting (NAS) and non-neoadjuvant setting (NNAS).

Design.—: A total of 99 LNs from 47 neoadjuvant patients and 108 LNs from 56 non-neoadjuvant patients were identified. Discordant cases were reviewed retrospectively to reveal the discrepancy reasons. Clinical, radiologic, and pathologic data were obtained from chart review and pathology CoPath database.

Results.—: The sensitivity, specificity, and negative predictive value of TP in NAS and NNAS were 34.2% versus 37.5%, 100% versus 100%, and 70.9% versus 90.2%, respectively. In NAS, discrepancy reasons were interpretation challenge due to lobular histotype, poor TP quality secondary to therapy-induced histomorphologic changes, and undersampling due to small tumor deposits (≤2 mm); the latter was the major reason in NNAS. More cases with macrometastasis were missed in NAS compared to NNAS (14 of 25 versus 1 of 10). The parameters associated with discrepancy were lobular histotype, histologic grade 2, ER positivity, HER2 negativity, multifocality, and pathologic tumor size greater than 10 mm in NAS; and lymphovascular space involvement and pathologic tumor size greater than 20 mm in NNAS.

Conclusions.—: In NAS, intraoperative TP alone should be used very cautiously owing to high false-negative rate of macrometastasis, especially for patients with invasive lobular carcinoma and known axillary LN metastasis before neoadjuvant therapy.