Purpose: Locoregional control is a prerequisite to cure primary breast cancer but the prediction of locoregional recurrence to guide further local therapy following neoadjuvant chemotherapy remains a challenge. The CPS + EG score was designed to predict distant recurrences. Here we examine its ability to predict both not only distant but also locoregional recurrences with respect to accuracy and clinical applicability.
Methods: Clinical data from 432 patients with primary breast cancer treated with neoadjuvant chemotherapy between 2003 and 2011 were prospectively collected. Using the Kaplan-Meier method we analyzed the risk of local and distant recurrences according to individual CPS + EG scores, stratified by type of surgery. Possible confounding of the relationship between recurrence risk and CPS + EG score by established risk factors was accounted for in multiple survival regression models. Additionally, we analyzed the performance of the CPS + EG score to predict isolated locoregional recurrence by censoring patients with prior or simultaneous distant metastases.
Results: 5-year locoregional recurrence-free survival was 90%, and 5-year distant metastases-free survival was 82%. The CPS + EG score stratified patients into six prognostic groups with distinct 5-year locoregional recurrence-free survival, ranging from 100 to 41% (p = 0.02) and 5-year distant metastases-free survival, ranging from 96 to 35% (p < 0.0001). 8 patients (17%) with CPS + EG scores ≥ 4 experienced locoregional recurrence-5 of them presented with simultaneous distant disease.
Conclusion: The CPS + EG score, originally designed to predict distant relapse, is also valuable for assessing local recurrence risks. Our data demonstrate that distant and locoregional recurrence risks are closely related. As prognosis of patients with high risk of locoregional failure based on CPS + EG is dominated by distant recurrences, escalating local therapies may have limited impact on overall prognosis.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1007/s10549-019-05314-9 | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!