Skip to main content
  • Oral Presentation
  • Published:

Molecular profiling of early breast cancer in relation to detection of micrometastases and outcome


Molecular profiling of breast cancer by DNA microarrays has been used to classify tumors into five distinct subclasses that show significant differences in clinical outcomes. Of these subclasses, the luminal subtype A is associated with a relatively good prognosis [13]. Detection of disseminated tumor cells (DTC) in bone marrow (BM) can independently predict future metastasis, which was confirmed in our study of 817 early BrCa patients [4].

Materials and methods

Fresh tumor samples were prospectively collected during primary surgery from 123 of these patients, for evaluation of the clinical significance of gene expression profiling and for comparison of tumor subtypes with DTC detection in BM. The BM samples were collected from iliac crests at primary surgery, followed by immunocytochemical staining (anti-cytokeratin mAbs) and morphology-guided screening for DTC. Gene expression patterns of the primary tumors were examined using 42,000 spot cDNA microarrays (Stanford Functional Genomics Facility). Data were analyzed by hierarchical clustering and were compared with our previously published breast tumor subclassifications. Data were further analyzed by supervised analysis methods (SAM, PAM).


The tumors were classified by gene expression analysis into luminal A (41%), luminal B (13%), ERBB2+ (17%), basal-like (14%) and normal-like (12%). The luminal A subtype showed high ER/PgR-positivity (98%), low ERBB2-positivity (4%) (assessed by IHC) and low frequency of TP53 mutations (6%). Luminal B, ERBB2+ and basal-like subtypes showed high frequencies of TP53 mutations (43%, 65%, and 82%, respectively), whereas the ER/PgR-positivity was 94%, 24% and 6%, respectively. Expression of the ERBB2 protein differed between these groups. At median 60 months follow-up, luminal A patients showed improved survival compared with patients within the other subtypes (P = 0.02, log rank), with BrCa death in 14% versus 29%, respectively. DTC in BM were detected in 23.7%. No particular subtype was associated with DTC, and no particular gene profile was associated with DTC status, as determined by SAM analysis. However, when we stratified the patients based on the molecular subtype, and first considered only the luminal A tumors, we identified 193 genes (FDR 23%) associated with high expression in tumors from patients with DTC. Moreover, a considerable number of patients with a luminal A type of tumor experienced systemic relapse of the disease (28%) and SAM analysis identified 147 genes associated with different expression patterns in tumors from relapsed patients versus disease-free patients.


This early BrCa study confirms the consistency of the gene expression profiles and their clinical implications. DTC detection can further distinguish the clinical outcome in patients with the luminal A subtype. The gene expression patterns in DTC-positive patients, and in all patients with systemic relapse, will be further explored.


  1. Perou CM, Sørlie T, Eisen MB, et al: Nature. 2000, 406: 747-752. 10.1038/35021093.

    Article  CAS  PubMed  Google Scholar 

  2. Sørlie T, Perou CM, Tibshirani R, et al: Proc Natl Acad Sci USA. 2001, 98: 10869-10874. 10.1073/pnas.191367098.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Sørlie T, Tibshiranhi R, Parker J, et al: Proc Natl Acad Sci USA. 2003, 100: 8418-8423. 10.1073/pnas.0932692100.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Wiedswang G, Borgen E, Kåresen R, et al: J Clin Oncol. 2003, 21: 3469-3478. 10.1200/JCO.2003.02.009.

    Article  CAS  PubMed  Google Scholar 

Download references

Author information

Authors and Affiliations


Rights and permissions

Reprints and permissions

About this article

Cite this article

Naume, B., Sørlie, T. Molecular profiling of early breast cancer in relation to detection of micrometastases and outcome. Breast Cancer Res 7 (Suppl 2), S.35 (2005).

Download citation

  • Published:

  • DOI: