Sentinel node biopsy in breast cancer : aspects on validation, diagnostics and lymphatic drainage pattern

Sammanfattning: Axillary lymph node status is the most important pathological determinant of prognosis in early breast cancer. Determination of axillary status is crucial in clinical decision-making. Total axillary lymphadenectomy is generally accepted as a reliable staging procedure. Routine axillary lymph node dissection (ALND) is being increasingly replaced by sentinel node biopsy (SNB) - detection and analysis of the first lymph node that drains the tumour. Immediate, complete ALND after SNB has confirmed that tumour-negative sentinel nodes accurately predict tumour-free axillary nodes in clinically node-negative breast cancer patients. This thesis validates the feasibility and accuracy of SNB in palpable and non-palpable breast cancer, and in women who have had an open biopsy procedure prior to SNB. It also discusses the short-term outcome of patients undergoing SNB as the only axillary procedure and showing a tumour-free sentinel node. Sentinel node biopsy was performed in stages T1 and T2 and clinically node-negative breast cancer patients. A confirmative axillary dissection was performed in paper I and the first part of paper III, whereas in the application phase of paper III, lymph node dissection was performed only in those patients who had metastases in the sentinel lymph node. The detection rate of SN was 90% in palpable tumours in the learning study (n=498 women; paper I), 95% in 57 women with non-palpable tumours and 96% in 75 women with an earlier breast biopsy (paper III). The false-negative rate was 11% in the learning study (paper I), 5.5% in the group of nonpalpable tumours, and 10% among those with an earlier open biopsy (paper III). In the application phase (paper III), 745 patients with non-palpable breast cancer and 86 patients with prior intervention were included. The detection rates were 95.3% (710/745) and 90.7% (78/86), respectively. During the follow-up time of almost 2 years, no axillary relapse was found among the 103 patients with a positive SLN that was followed by ALND. Two axillary relapses were found among the 607 patients with a negative SNB. No axillary relapse was found among women who had had an open biopsy procedure prior to SNB (paper III). All sentinel nodes were submitted for histopathological processing using frozen-section examination with hematoxylin-eosin staining (H&E). In a separate study, H&E was compared with immunohistochemistry with cytokeratin antibodies (MC), and imprint cytology (IC) (paper II). The overall sensitivity of intraoperative frozen section examination with H&E staining was 72.3%, with IHC 72.3% and with IC 48.9%. The accuracy of the three methods was 87.3%, 87.3% and 76.5%, respectively. Combining intraoperative frozen section with H&E staining and IHC raised the sensitivity to 80.9%, whereas the addition of intraoperative IC examination did not affect the results. The sensitivity of intraoperative frozen section for micrometastases was 35% with H&E staining, 45% with MC and 55% with IC (paper II). In a study including 30 patients operated with SNB alone and 30 patients with SNB and axillary clearance (ALND), a comparison was made of the clinical outcome, lymphoscintigraphy of the arm, arm volume, skin circulation and skin temperature, 2-3 years after radiotherapy (paper IV). None of die 30 patients operated with SNB showed any clinical manifestation of lymph oedema. Of the 30 patients operated with ALND, 6 (20%) had clinical lymph oedema, defined as an arm volume increase of more than 10% in the affected arm compared to the nonoperated arm (paper IV). Conclusion: Sentinel node biopsy is feasible and safe in palpable and non-palpable breast cancer and after previous breast biopsy. Sentinel node biopsy can accurately predict the nodal status of the axilla. The intraoperative frozen section analysis of SNB with H&E staining showed an acceptable sensitivity in detecting macrometastases in the lymph node. Lymph drainage in the operated arm and morbidity both seemed to be less affected by SNB than by ALND.