Abstract 1521
Background
Patients with metastatic breast cancer (MBC) is a heterogeneous group with different survival outcomes. The current "catch-all" M1 category of breast cancer cannot be used for accurate prognosis prediction.
Methods
Patients with de novo MBC were identified using the Surveillance, Epidemiology, and End Results (SEER) database and divided randomly into the training and validation sets. The Fine and Gray's competing risks model was developed to identify the variables associated with increased breast cancer-specific death (BCSD) in the training set. Cumulative incidence curves were estimated and compared using Gray's test. And the M1 subdivisions system was established based on the independent prognostic factors for BCSD.
Results
Multivariate analysis showed the involvement of brain or liver and the number of metastatic organs were independent prognostic factors for BCSD. Therefore, we subdivided the M1 stage into three categories: M1a, involvement of single organ but no brain or liver; M1b, liver metastasis only or involvement of multiple organs but no brain or liver; M1c, involvement of multiple organs including liver but no brain, or brain involvement with or without liver, irrespective of the number of metastatic organs (M1b vs M1a, subdistribution hazard ratio [SHR] 1.45, 95% CI 1.28 - 1.65; M1c vs M1a, SHR 2.47, 95% CI 2.22 - 2.75; M1c vs M1b, SHR 1.67, 95% CI 1.47 - 1.90). The dose-response risk estimation was also observed in the validation and whole sets. Primary tumor surgery decreased from 43.2% in 2010 to 29.8% in 2014, with similar patterns seen in all M1 subdivisions. And patients of M1a benefited most from primary tumor surgery (M1a: SHR 0.55, 95% CI 0.49 - 0.60; M1b: SHR 0.71, 95% CI 0.60 - 0.82; M1c: SHR 0.63, 95% CI 0.55 - 0.72) in the adjusted competing risks model. Three-year cancer-specific mortality cumulative rates.Table: 355P
| Training set | Validation set | Whole set | ||||
|---|---|---|---|---|---|---|
| M1 subdivision | Incidence rate (%) | 95% CI (%) | Incidence rate (%) | 95% CI (%) | Incidence rate (%) | 95% CI (%) |
| M1a | 42.4 | 39.8 - 44.9 | 44.1 | 41.6 - 46.7 | 43.2 | 41.4 - 45.1 |
| M1b | 53.7 | 49.1 - 58.0 | 50.6 | 46.3 - 54.8 | 52.1 | 48.9 - 55.1 |
| M1c | 72.1 | 68.3 - 75.6 | 70.4 | 66.4 - 74.0 | 71.3 | 68.6 - 73.8 |
| P | < .001 | < .001 | < .001 |
Conclusions
The M1 subdivisions system can better guide the prognosis prediction and treatment planning in patients with de novo MBC.
Clinical trial identification
Legal entity responsible for the study
Ethics Committee, Shanghai Ruijin Hospital, Shanghai Jiao Tong University, College of Medicine, Shanghai, CN.
Funding
Has not received any funding.
Editorial Acknowledgement
Disclosure
All authors have declared no conflicts of interest.