Brain Metastases from Breast Cancer: Recursive Partitioning Analysis of Prognostic Factors Including Molecular Subtypes and Treatment

Reviewer: Geoffrey Geiger, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: October 31, 2010

Authors: R. Le Scodan, C. Massard, L. Jouanneau, F. Coussy, M. Gutierrez, Y. Kirova, A. Labib, E. Mouret-Fourme
Institution: Institut Curie, Hopital Rene Huguenin, Saint Cloud, France; Institut Gustave Roussy, Villejuif, France; Institut Curie, Site Ulm, Paris, France


  • Approximately 20-40% of all cancer patients will develop intracranial metastases during their disease course.
  • Solid cancers of the lung and breast are two of the most common malignancies with proclivity for spread to the brain and approximately 10-15% of breast cancer patients will ultimately develop brain metastases.
    • These patients have a median survival of approximately 5 months.
  • The RTOG has published a recursive partitioning analysis (RPA) looking at three classes of patients with brain metastases:
    • Class I: KPS 70-100, age <65, metastases to brain only
      • Median survival 7.1 months
    • Class II: KPS 70-100, age > 65 OR metastases to elsewhere sites.
      • Median survival 4.2 months
    • Class III: KPS <70
      • Median survival 2.3 months
  • Treatment recommendations for patients with brain metastases can be broken down by RPA class, with whole brain radiotherapy (WBRT) alone suggested for patients with RPA class III.
    • For patients with RPA class I-II disease and a single lesion, whole brain radiation plus stereotactic radiosurgery (SRS) or resection plus whole brain radiation is suggested.
    • For patients with two or more metastases, WBRT or WBRT+SRS is suggested.
  • This investigation was designed to determine whether a disease-specific RPA analysis could be developed to prognosticate survival in metastatic breast cancer (mBC) patients using molecular subtypes and treatment parameters.

Materials and Methods

  • Investigators retrospectively examined a patient population of 190 patients, of which 130 were ultimately selected for use in developing an RPA analysis for patients treated between January 1998 and April 2006.
  • All patients had 3+ metastases and were treated with WBRT (with or without SRS).
  • Outcomes were utilized to stratify patients according to various disease and molecular characteristics to develop an RPA analysis.
  • The final RPA divisions were grouped according to their ultimate survival times.
  • Patients were ultimately analyzed utilizing seven prognostic factors: 1) Performance status; 2) Age; 3) Trastuzumab-based therapy for HER2 overexpressing tumors; 4) Triple-negative phenotype; 5) Scarff-Bloom-Richardson grade; 6) Serum LDH level and 7) Lymphocyte count at mBC diagnosis.


  • Overall survival at 6 months was 54.9%, and 35.8% at 1 year. Median overall survival was 7.43 months.
  • When patients were broken down by RTOG RPA, 82 were patients classified as class I-II and had a median survival of 9.63 months. Patients classified as RTOG RPA class III had a median overall survival of 3.52 months (p=0.0001)
  • The majority of patients were treated with 30 Gy delivered in 10 fractions.
  • The best median survival (median 19.5 months) was observed in patients with HER2-overexpressing tumors treated with trastuzumab-based therapy.
  • The worst survival (median 3.5 months) was observed in patients with HER2-overexpressing tumors not treated with trastuzumab and with lymphopenia at mBC diagnosis, or with a KPS <70 and older than 50 years, or in patients with KPS <70 with triple negative (ER-, PR-, HER2-) tumors.
  • All other patients, when considered as a single group had a median survival of 12.49 months.
  • Survival analysis was performed using the Kaplan-Meier method and a log-rank test was used to confirm significant differences between groups.

Author’s Conclusions

  • This study is the first to apply a disease-specific RPA analysis to mBC patients, and to explain survival differences between groups using molecular subtypes and treatment parameters.
  • HER2 overexpressing tumors in patients treated with trastuzumab-based therapy had the highest survival of any analyzed group with a 1-year survival of 60% and a median survival of nearly 20 months.
  • Authors separated patients into three separate groups and the prognostic score successfully predicted the outcome of a heterogeneous group of patients with mBC.
  • This information may be useful to tailor the therapy for subgroups of patients and to define homogeneous cohorts for retrospective comparison and prospective randomized trials.

Clinical Implications

  • Brain metastases are common in patients with mBC and when considered as a group, only about 20% of these patients are alive 1 year following the discovery of brain metastases.
    • Patients with several brain metastases, such as the patient population in this study, who were not candidates for radical surgery or stereotactic radiosurgery, have improved survival with the addition of WBRT.
    • It would be useful to be able to prognosticate patients who are only eligible for WBRT according to breast cancer specific criteria, which was the aim of this investigation.
  • The data demonstrate that patients who received trastuzumab-based therapies for mBC in the context of WBRT for multiple brain metastases had significantly improved outcomes when compared the population as a whole.
  • Patients with HER2 overexpressing tumors who did not receive trastuzumab-based therapy had a much worse prognosis, as did patients with lymphopenia at mBC diagnosis or those patients with a KPS <70 who were older than 50 years or patients with KPS <70 with triple negative tumors.
  • There are several limitations of this study:
    • No data were given as to when trastuzumab-based therapy was started (i.e., in the adjuvant or metastatic setting) or when HER2 testing was performed (i.e., at the time of diagnosis or at the time of the diagnosis of mBC). There was no information presented regarding methods by which HER2 status was assessed (i.e., IHC vs. FISH testing).
    • The data regarding lymphopenia is likely confounded by the timing of the blood draw (i.e., was the patient lymphopenic at the time of mBC diagnosis when they were likely off systemic therapy or at the time when brain metastases were diagnosed).
    • Disease burden within the CNS and outside of the CNS was not examined. Furthermore, the number of brain metastases was not included in the RPA analysis. All of these factors would be expected to influence patient outcomes.
    • Authors did not report data regarding neurological death.
    • No data was given regarding ER/PR status in this patient population.
    • There was no breakdown regarding patients that may have received an SRS boost, or by dose during WBRT.
  • Nonetheless, although all the caveats of a retrospective analysis apply here, this investigation does provide some disease-specific criteria to use for the prognostication of mBC patients with brain metastases. These data will need to be validated in the prospective setting before being more widely utilized.