(P 210) RADIATION-ASSOCIATED ANGIOSARCOMA OF THE BREAST (RAASB): A NOMOGRAM FOR PREDICTING RECURRENCE AND PROGNOSIS

Objective: To establish the nomogram for predicting recurrence risk and prognosis for patients with radiation-associated angiosarcoma of the breast (RAASB).

Methods: Cases published before June 2024 were comprehensively screened in Embase and Medline. Clinicopathological features, including age at time of diagnosis of RAASB, interval between the diagnosis of breast cancer and diagnosis of RAASB, primary tumor grade, AJCC stage of primary tumor, adjuvant therapy for the breast cancer, RAASB size, RAASB grade, 1st-line treatment for RAASB, local recurrence-free interval (LRFI), and overall survival (OS) were recorded. Survival outcomes were compared using Kaplan-Meier analysis with log-rank tests. Prognostic factors were identified using Cox regression analysis and further visualized with nomogram to predict 1-year, 2-year, and 5-year LRFI as well as OS. The model was externally validated by the cohort consisting of patients diagnosed with RAASB since 2002 at Mayo Clinic in Rochester, Minnesota.

Results: A total of 165 patients diagnosed with breast cancer since 1986 were enrolled as the training cohort and 11 patients were identified as the validation cohort. At the time of diagnosis of RAASB, the median age of patients was 67 (29-86) years; the median interval was 72 months. The median LRFI (mLRFI) was 32 months, and the median OS (mOS) was 91 months. Radiation therapy significantly prolonged LRFI for patients with RAASB (mLRFI 32 months vs. 13 months, P=0.005), while no significant difference was detected regarding chemotherapy. However, neither radiation therapy nor chemotherapy as 1st-line treatment exerted independent effects on OS of patients with RAASB. Prognostic indicators for LRFI include age (P=0.01), interval between RAASB and breast cancer diagnoses (P=0.008), adjuvant chemotherapy for breast cancer (P=0.019), RAASB size (P=0.006), and 1st-line radiation therapy (P=0.011). Nomogram for predicting LRFI showed reliable accuracy with the C-index of 0.77 (0.73-0.80) in the training cohort and 0.82 (0.71-0.93) in the validation cohort. The prognostic factors for OS consisted of adjuvant chemotherapy (P=0.031), RAASB size (P=0.016), and RAASB grade (P=0.044). The C-index of nomogram for OS was 0.73 (0.67-0.78) in the training cohort and 0.91 (0.84-0.98) in the validation cohort, suggesting clinically meaningful predictive power.

Conclusion: This study established and validated a prediction model for patients with secondary angiosarcoma developed after radiation treatment of breast cancer. Understanding of individualized risk in conjunction with patient goals of care may help guide treatment decisions and ongoing monitoring. Future studies are warranted to further validate this prediction model for patients with RAASB.