MADRID, Spain, October 27, 2026 — Burning Rock Biotech Limited (“Burning Rock”) and its clinical collaborators from Jiangsu Province Hospital and the Affiliated Hospital of Qingdao University presented clinical validation results (Poster 2458P) at the European Society for Medical Oncology (ESMO) Congress 2026, exploring STELLA’s potential to support the differentiation of benign and malignant breast lesions as an adjunct to breast imaging.
Accurate differentiation of benign and malignant breast lesions is essential for personalized clinical management. To evaluate the role of liquid biopsy in this setting, Burning Rock and its clinical collaborators conducted the prospective PERCEIVE-BREAST study. The study used Burning Rock’s SPIRAL platform to detect methylation signals and applied STELLA, a tumor-agnostic, DNA methylation-based algorithm, to quantify circulating tumor DNA (ctDNA) levels and evaluate its clinical performance in supporting the differential diagnosis of breast lesions.
Blood samples were prospectively collected before any anti-tumor therapy. The analysis included 516 patients: 395 with invasive breast cancer, 15 with in situ carcinoma, and 106 with benign breast lesions. The study evaluated overall sensitivity and specificity, with further analyses by clinical stage, molecular subtype, and imaging classification.
STELLA achieved 65.9% overall sensitivity and 97.2% specificity. Sensitivity increased with disease stage, reaching 32.7%, 77.4%, 85.3%, and 100.0% in stages I, II, III, and IV, respectively. STELLA also detected 5 of 15 in situ carcinomas, corresponding to a sensitivity of 33.3%. Across molecular subtypes, sensitivity was highest in HR+/HER2+ disease at 89.9%, followed by HR−/HER2− at 81.4%, HR−/HER2+ at 79.5%, and HR+/HER2− at 51.6%.
Among patients classified as BI-RADS category 4 on ultrasound, mammography, or MRI, STELLA demonstrated sensitivities of 55.5%, 59.7%, and 63.3%, respectively, with corresponding specificities of 98.3%, 100.0%, and 95.0%.
These findings highlight STELLA’s potential as an adjunctive tool for differentiating benign and malignant breast lesions. Its high specificity may support its use alongside conventional imaging to help identify a subset of BI-RADS category 4 patients with a high probability of malignancy. Given the lower tumor burden in early-stage disease, a negative result alone cannot rule out malignancy; assessment should integrate follow-up imaging and pathological examination. The research team will also incorporate follow-up data to further explore STELLA’s potential prognostic value in patients with breast cancer.



