Characterization of Oligometastatic Disease in a Real-World Nationwide Cohort of 3447 Patients With de Novo Metastatic Breast Cancer

Abstract

Background: Observational studies in metastatic breast cancer (MBC) show that long-term overall survival (OS) is associated with limited tumor burden, or oligo-MBC (OMBC). However, a uniform definition of OMBC is lacking. In this real-world nationwide cohort, we aimed to define the optimal OMBC threshold and factors associated with survival in patients with OMBC.

Methods: 3535 patients aged younger than 80 years at diagnosis of de novo MBC in the Netherlands between January 2000 and December 2007 were included. Detailed clinical, therapy, and outcome data were collected from medical records of a sample of the patients. Using inverse-sampling-probability weighting, the analysis cohort (n = 3447) was constructed. We assessed OS according to number of metastases at diagnosis to determine the optimal OMBC threshold. Next, we applied Cox regression models with inverse-sampling-probability weighting to study associations with OS and progression-free survival in OMBC. All statistical tests were 2-sided.

Results: Compared with more than 5 distant metastases, adjusted hazard ratios for OS (with 95% confidence interval [CI] based on robust standard errors) for 1, 2-3, and 4-5 metastases were 0.70 (95% CI = 0.52 to 0.96), 0.63 (95% CI = 0.45 to 0.89), and 0.91 (95% CI = 0.61 to 1.37), respectively. Ten-year OS estimates for patients with no more than 3 vs more than 3 metastases were 14.9% and 3.4% (P < .001). In multivariable analyses, premenopausal andperimenopausal status, absence of lung metastases, and local therapy of metastases (surgery and/or radiotherapy) added to systemic therapy were statistically significantly associated with better OS and progression-free survival in OMBC, independent of local therapy of the primary tumor.

Conclusion: OMBC defined as MBC limited to 1-3 metastases was associated with favorable OS. In OMBC, local therapy of metastases was associated with better OS, particularly if patients were premenopausal or perimenopausal without lung metastases.

Figure 1.

Schematic presentation of matched and weighted cohort. A) The population shown represents the complete population of patients diagnosed with metastatic breast cancer between 2000 and 2007 in the Netherlands. Dots are coded to illustrate the distribution of clinical characteristics, such as estrogen receptor status and age at diagnosis. B) Shows 5.9% of patients, who are alive more than 10 years since diagnosis. These patients have a different distribution of clinical characteristics. To allow exploration of biomarkers, we matched known clinical characteristics, which were ER status, age at diagnosis, and year of diagnosis, with a sample of patients not surviving 10 years (D). To evaluate the effect of clinical characteristics and other known clinical characteristics, we calculated inverse-probability-weights based on the known distribution of ER, age groups at diagnosis, and year of diagnosis in the complete population (A) and the sample (D) to reconstruct the whole population (C). Population (B) and (C) are used for the analyses. IPW = inverse probability weight; MBC = metastatic breast cancer.

Figure 2.

CONSORT diagram. Thebox indicates the analysis cohort of patients with OMBC (n=517). For all sampled patients surviving less than 10 years, the inverse probability sample weight was calculated based on year of diagnosis, age categories (20-39, 40-49, 50-59, 60-69, and 70-79 years), and ER status. Patients alive 10 years or more had a sample weight of 1. MBC = metastatic breast cancer.

Figure 3.

Between 2000 and 2007, on average, 5.9% of 3535 patients with de novo MBC survived 10 years or longer. The x-axis shows the year of diagnosis; the left y-axis represents number of patients diagnosed with de novo MBC in the Netherlands younger than 80 years (bars); the right y-axis represents the percentage of patients surviving at least 10 years (darksolid line). MBC = metastatic breast cancer.

Figure 4.

The 10-year OS estimate of patients with no more than 3 metastases is 14.9% vs 3.4% for patients with more than 3 metastases. Univariable hazard ratio is reported, and the 95% confidence interval is based on robust standard errors. Numbers of patients are based on the weighted cohort of patients. Number of metastases was known for 3140 patients. CI = confidence interval; HR = hazard ratio.