A 3-Dimensional CT-Based Analysis of Dosimetric Coverage on the Patterns of Locoregional Recurrence in High-Risk Breast Cancer Patients Treated with Regional Nodal Irradiation

Abstract

Maximizing dose to regions-at-risk is counterbalanced by efforts to minimize dose to adjacent organs during the treatment planning of locally advanced breast cancer patients. Our goal was to evaluate the association of dosimetric coverage of the target tissues (breast, chest wall, regional nodes) with the incidence of locoregional recurrence (LRR) in breast cancer patients who had been previously treated with RNI. From an institutional database, we identified 57 patients treated for Stage II-III breast cancer between 2000-2014 at single center who developed an isolated LRR following surgery and adjuvant radiation to the breast/chest wall and regional nodes. Eligibility criteria included receipt of conformal RT to the lymph nodes, available DVH data, and radiographic identification of recurrent tumor on imaging scans obtained within 3 months of LRR. Patients with prior distant metastases, recurrence without correlative imaging and prior metastases were excluded. In order to determine the dose coverage in the area of recurrence, we fused the diagnostic chest CT demonstrating recurrence to the original treatment planning CT to create a gross tumor volume-LRR (GTV-LRR). The mean doses to the breast or chest wall, supraclavicular (SCV), Level 1-3 axillary (Ax) and internal mammary nodes (IMN), heart dose (MHD), left anterior descending dose and lung (MLD) were evaluated by DVH. The log-rank test evaluated the association between target coverage and time to the development of LRR. Cox multivariate analysis was adjusted for receptor subtype, receipt of systemic therapy and pathologic stage. A total of 48% of the cohort was comprised of in-breast or chest wall recurrences (BR) and 52% isolated nodal recurrences (NR). Among the NR group, 38% patients received a mean <30 Gy to the GTV-LRR and 62% received mean >30 Gy to the GTV-LRR. The majority of GTV-LRRs in the NR group were in the IMN chain (38%) and axilla (38%), followed by SCV (24%). In the BR group, the association between GTV-LRR and time to LRR was not significant. In contrast, in the NR group, the median time to LRR in patients who received mean<30 Gy GTV-LRR was 7 months , compared to the mean >30 Gy GTV-LRR cohort who failed at a median of 21 months (p=0.03). The mean heart dose in the <30 Gy group was significantly lower, compared to the >30 Gy cohort (p=0.02), indicating that sparing of the heart was prioritized over coverage of target tissues. In this CT-based mapping analysis of LRRs in breast cancer patients previously treated with RNI, the time to development of a nodal recurrence was prolonged by 3-fold when the area had received mean<30 Gy relative to those who had received >30 Gy to the area of eventual recurrence. This study provides proof-of-concept that dosimetric coverage of nodal areas-at-risk should be prioritized in high-risk breast cancer patients receiving radiotherapy.