Can Molecular Profiling Identify ER-positive Patients Destined for Relapse?

Researchers improve understanding of the biology for early, vs later, recurrence.

Get Permission

Estrogen receptor (ER)-positive breast cancer often recurs years after the initial diagnosis, and understanding the patterns of timing regarding relapse could identify patients needing more aggressive treatment. At the 2011 San Antonio Breast Cancer Symposium, several investigative teams reported progress in defining these patterns and risk categories.

Robust Molecular Differences

3.3s.11_liu.jpg“Our work provides clear evidence that robust molecular differences exist between ER-positive breast cancers that recur early vs much later, despite adjuvant tamoxifen,” said Minetta C. Liu, MD, of the Georgetown Lombardi Comprehensive Cancer Center, Washington, DC.

Dr. Liu and colleagues analyzed gene-expression data from breast tumor biopsies, and then correlated them with the development of distant metastases.1 What emerged was a 91-gene classifier that reliably separates early recurrences (distant relapse ≤ 3 years from diagnosis) from late recurrences (≥ 10 years).

The gene classifier was derived from pretreatment tumor biopsies collected at the Edinburgh Breakthrough Research Unit between 1982 and 1990 from 111 patients with stage I to III ER-positive breast cancer who received adjuvant tamoxifen. The optimized classifier was validated on a large independent data set and found to have high levels of accuracy, specificity, sensitivity, positive predictive value, and negative predictive value, Dr. Liu noted. In addition, patients predicted to experience an early recurrence had a greater than threefold risk of dying in both the training dataset (P < .0001) and the validation dataset (P = .0004).

3.3s.12_quote.jpgA novel computational procedure was developed to integrate gene-expression data with protein–protein interaction data and thus create a statistical network model of signaling. The majority of genes identified through network modeling were related to apoptosis and proliferation. For example, tumors from patients with an early recurrence had increased expression of CALM1, CALM2, CALM3, SRC, CDK1, and MAPK1. For late recurrences, increased tumor expression of ESR1, ESR2, EGFR, BCL2, and AR was observed.

“Reliable prediction of early treatment failure may identify patients who require agents beyond endocrine therapy to prevent the early onset of distant metastases,” Dr. Liu said. “Exploiting the molecular differences between early vs late recurrences may also guide the development of effective novel drug combinations in this group.”

She added that her group’s goals extend beyond development of a score predictive of long-term outcomes. “We want to understand the underlying mechanisms for the development of early vs late breast cancer recurrences and develop ways to convert them to never recurrences,” she said.

International Investigation

An international study asked similar questions and discovered that patients could be risk-stratified for recurrence through a measure of genes related to proliferation plus those related to estrogen receptor expression.2

The study used three public gene-expression datasets involving 606 ER-positive patients who received tamoxifen. A measure of proliferation was calculated as the average expression of 12 mitotic kinases, which investigators called a mitotic kinase score (MKS). A four-gene estrogen-related score (ERS) was adopted from the ER gene subset of the Oncotype DX tool.

Four groups were evaluated: low MKS/high ERS; low MKS/low ERS; high MKS/high ERS; and high MKS/low ERS. Patterns of distant recurrences were assessed in three time intervals: upfront (0–2.5 years), after a switch to an aromatase inhibitor (2.5–5 years), and years after treatment (5–10 years).

The majority of relapses observed within the first 2.5 years occurred in patients with high proliferation and low estrogen-related scores. “These cancers are enriched in tumors intrinsically resistant to tamoxifen, and possibly to aromatase inhibitors,” said Giampaolo Bianchini, MD, of the Fondazione Centro San Raffaele del Monte Tabor in Milan, Italy.

Mitotic kinase score and estrogen-related score were both prognostic—independently and as a combined marker. High proliferation/low estrogen-related expression was associated with the worst outcomes, and low proliferation/high estrogen-related expression had the best outcomes. “But importantly, the odds ratios were not constant over time,” Dr. Bianchini said.

Time Patterns

Analysis of the time pattern of relapse by the biomarker groups at three time points showed unique risk profiles according to the MKS/ERS gene patterns. For example, compared with low MKS/high ERS tumors, those with high MKS/low ERS had an almost 11-fold increased risk for recurrence within 2.5 years, but no increased risk at 5 to 10 years.

“Low proliferation/high ERS tumors had an excellent prognosis, even after stopping tamoxifen,” he reported. “But high risk of recurrences after 5 years of tamoxifen was present in tumors with high MKS/high ERS. These patients may be the best candidates for extended endocrine therapy after 5 years of tamoxifen. High risk of late relapse after 5 years was also present in the low MKS/low ERS group, so we wonder if there is a role for aromatase inhibitors in this group as well.”■

Disclosure: Drs. Liu and Bianchini reported no potential conflicts of interest.

Expert Point of View #1: Can Molecular Profiling Identify ER-positive Patients Destined for Relapse?

Expert Point of View #2: Can Molecular Profiling Identify ER-positive Patients Destined for Relapse?


1. Liu MC, Dixon JM, Xuan JJ, et al: Molecular signaling distinguishes early ER-positive breast cancer recurrences despite adjuvant tamoxifen. 2011 San Antonio Breast Cancer Symposium. Abstract S1-8. Presented December 7, 2011.

2. Bianchini G, Pusztai L, Iwamoto T, et al: Molecular tumor characteristics influence adjuvant endocrine treatment outcome. 2011 San Antonio Breast Cancer Symposium. Abstract S1-7. Presented December 7, 2011.

Related Articles

Expert Point of View: Can Molecular Profiling Identify ER-positive Patients Destined for Relapse?

Eastern Cooperative Oncology Group (ECOG) trials have demonstrated that recurrence patterns are different for estrogen receptor (ER)-positive vs ER-negative breast cancer. While ER-negative disease carries a high risk for early recurrence—peaking around year 1.5 from diagnosis—risk steadily...

Expert Point of View #2: Can Molecular Profiling Identify ER-positive Patients Destined for Relapse?

Understanding the different relapse patterns within estrogen receptor (ER)-positive breast cancer is important for guiding treatment decisions, said Laura Esserman, MD, of the University of California, San Francisco, Medical Center. Women who fall into the two separate categories may have different ...