BACKGROUND: We implemented selective use of frozen section (FS) to optimize accuracy and cost control in the intraoperative diagnosis of sentinel lymph node (SLN) in patients with breast cancer, guided by the Memorial Sloan Kettering Cancer Center (MSKCC) nodal metastasis risk prediction nomogram. METHODS: Surgical pathology records were reviewed, examining 2582 consecutive biopsies from 2552 patients with breast cancer to compare intraoperative FS diagnoses with postoperative final reports. We calculated sensitivity, specificity, and false-negative rates (FNRs) for various MSKCC risk levels, also analyzing axillary reoperation rates, with and without FS, and the number needed to treat (NNT) to avoid separate axillary lymph node dissection. RESULTS: The sensitivity, specificity, and FNR of FS were 84.7%, 99.9%, and 15.3%, respectively. FNR and MSKCC risk level negatively correlated (r = -0.86; P = .002). Axillary reoperation rate significantly declined if FS was done (FS: 4.0%; no FS: 36.4%; P = .002). In grouping patients by quartile of MSKCC risk, axillary reoperation rates were 16.7%, 25.1%, 38.7%, and 58.7% without FS, compared with 4.3%, 3.2%, 5.6%, 3.3% with FS and NNT correspondingly fell from 8.1 to 4.6, 3.0, and 1.8. CONCLUSIONS: A stratified decision-making algorithm based on the MSKCC risk prediction model improved the effectiveness of FS during SLN biopsy to avoid axillary reoperation.
BACKGROUND: We implemented selective use of frozen section (FS) to optimize accuracy and cost control in the intraoperative diagnosis of sentinel lymph node (SLN) in patients with breast cancer, guided by the Memorial Sloan Kettering Cancer Center (MSKCC) nodal metastasis risk prediction nomogram. METHODS: Surgical pathology records were reviewed, examining 2582 consecutive biopsies from 2552 patients with breast cancer to compare intraoperative FS diagnoses with postoperative final reports. We calculated sensitivity, specificity, and false-negative rates (FNRs) for various MSKCC risk levels, also analyzing axillary reoperation rates, with and without FS, and the number needed to treat (NNT) to avoid separate axillary lymph node dissection. RESULTS: The sensitivity, specificity, and FNR of FS were 84.7%, 99.9%, and 15.3%, respectively. FNR and MSKCC risk level negatively correlated (r = -0.86; P = .002). Axillary reoperation rate significantly declined if FS was done (FS: 4.0%; no FS: 36.4%; P = .002). In grouping patients by quartile of MSKCC risk, axillary reoperation rates were 16.7%, 25.1%, 38.7%, and 58.7% without FS, compared with 4.3%, 3.2%, 5.6%, 3.3% with FS and NNT correspondingly fell from 8.1 to 4.6, 3.0, and 1.8. CONCLUSIONS: A stratified decision-making algorithm based on the MSKCC risk prediction model improved the effectiveness of FS during SLN biopsy to avoid axillary reoperation.