4MO - Machine learning intratumoral and peritumoral magnetic resonance imaging radiomics for predicting disease-free survival in patients with early-stage breast cancer (RBC-01 Study)

Background

There are no satisfying ways to distinguish high- from low-risk patients with early-stage breast cancer. We aimed to develop a MRI radiomic-based signature for predicting prognosis and discriminating of high-risk relapse patients with different molecular subtypes (RBC-01 study).

Methods

Machine learning intratumoral and peritumoral radiomics to develop the radiomic signature for DFS prediction in 799 patients from Sun Yat-sen Memorial Hospital and Sun Yat-sen University Cancer Center (training cohort). Clinical-radiomic nomogram was constructed by integrating radiomic signature with significant clinical risk factors. The performance of the model was validated in prospective phase III trials [NCT01503905] (internal validation cohort, n=105), and Shunde Hospital of Southern Medical University and Tungwah Hospital of Sun Yat-Sen University (external validation cohort, n=180).

Results

In the training cohort, the radiomic signature comprising intratumoral and peritumoral features showed an improved performance, with 1-, 2-, 3-year AUCs of 0.97, 0.95, and 0.98 over intratumoral or peritumoral radiomics alone. The clinical-radiomic nomogram achieved the highest 1-, 2-, 3-year AUCs of 0.97, 0.96, and 0.98, it was also found to be significantly associated with DFS (HR 0.027, 95% CI 0.010-0.077, p<0.001). The prognostic value was validated in the internal and external cohorts. The clinical-radiomic nomogram could also discriminate high- from low-risk patients in different molecular subtypes (P<0.001 for Luminal A; P<0.001 for Luminal B; P=0.007 for HER2+; P<0.001 for TNBC). Neoadjuvant chemotherapy improved DFS compared with patients who received adjuvant chemotherapy (P=0.048), among high-risk patients of Luminal subtype. No significance was observed between neoadjuvant chemotherapy and adjuvant chemotherapy in patient with low-risk (P=0.400).

Conclusions

The clinical-radiomic nomogram we developed which shows the potential to be served as a convenient tool for DFS prediction in patients with early-stage breast cancer and identify patients who might benefit from neoadjuvant chemotherapy.

Clinical trial identification

NCT04003558; ChiCTR1900024020.

Editorial acknowledgement

Legal entity responsible for the study

Sun Yat-sen Memorial Hospital, Sun Yat-sen University.

Funding

National Natural Science Foundation of China;National Major Science and Technology Projects of China; Medical Artificial Intelligence Project of Sun Yat-Sen Memorial Hospital; Natural Science Foundation of Guangdong Province; Guangzhou Science and Technology Major Program; Sun Yat-Sen University Clinical Research 5010 Program; Sun Yat-Sen Clinical Research Cultivating Program; Guangdong Science and Technology Department.

Disclosure

All authors have declared no conflicts of interest.