Analysis of Patterns of Lymph Node Metastases in Primary and Recurrent Breast Cancer by Evaluation of FDG / PET-CT Image Datasets

Abstract

The understanding of lymphogenic metastasis pattern is crucial in the adjuvant radiotherapy of breast cancer patients. An exact determination of the regions in which lymph node metastases (LN) occur is necessary to ensure a therapeutic effect when lymph node levels (LNL) are part of the target volume. The aim of this study was to localize loco-regional LN using FDG-PET datasets in a large cohort of patients and to evaluate the existing RTOG contouring guidelines. A total of 1,326 patients with breast cancer underwent FDG/PET-CT for staging purposes in our institute between 01/2014 and 03/2017. 235 of these patients with FDG/PET-CT positive loco-regional LN were included in the study. The FDG/PET-CT images were imported into the planning software and each PET positive LN was contoured separately. Patients were classified in four groups: primary vs. recurrent breast cancer; M1 vs. M0. By means of non-rigid registration all contoured structures were transferred to a patient with standard anatomy. Thus a 3D LN atlas was created including all LN metastases. LNL (axillary, internal mammary, supraclavicular) were contoured in the standard patient according to RTOG guidelines. We assed if the LN were “in field”, “marginal” or “out of field” the RTOG LNL. Bulky nodes (>25cm3) were excluded from this study (i.e. 13 LN). A total of 585 LN metastases were detected in the described collective. The mean volume of the lymph nodes was 1.9±3.3 cm3 with an average diameter of 1.3±0.7 cm. Most lymph nodes were localized in level I (n=311; 53.2%) followed by the supraclavicular region (n=79; 13.5 %), level II (n=59; 10.1 %), level III (n=59; 10.1%) and the internal mammary region (n=54; 9.2 %). 23 LN were located in the breast tissue itself (intramammary). In M0, primary patients level I, II and III were the predominant localisation, whereas most of the lymph nodes in the internal mammary region and supraclavicular region were observed in recurrent breast cancer with distant metastases. 337 (57.6 %) of the LN were completely within the RTOG LNL boundaries. In 164 (28.1 %) cases the LN were only partly in the defined lymph node levels. 82 lymph nodes (14.2 %) were completely outside the defined RTOG LNL. In this study, a comprehensive lymph node atlas was created using PET data sets. Our data indicate that the lymph node localizations are dependent on the stage of the disease (recurrent vs. primary; M0 vs. M1) and that recurrent lymph node metastases occur more frequently in the supraclavicular and internal mammary region compared to the primary situation. In many cases the RTOG guidelines do not reliable reflect the involved lymph node regions.