Highlights of the San Antonio Breast Symposium Part II

This article has been archived.
Please use for reference only.

Julia Draznin Maltzman, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: February 15, 2004

Predicting disease recurrence

There has been much research dedicated to identifying markers, or patient characteristics, that would be useful in predicting the risk of recurrence from breast cancer following initial surgery. Traditionally, physicians used prognostic factors such as tumor size, number of involved lymph nodes, tumor grade, invasion into the lymphatic system, and receptor positivity to predict the chances of recurrence. Physicians also use this information to help decide on the type of therapy to offer or even whether or not to offer therapy at all. In general, when the risks of recurrence are greater than the risks of therapy, most physicians would support treatment. Of course, each case must be considered individually to take into account all other factors.

Unfortunately, even with all the currently available medical knowledge, our methods for predicting outcomes are still quite crude and simple. As a result, we may over treat some patients while under treating others. If we could know in advance which patients? are likely to recur, we would offer them more aggressive therapy while, avoiding exposure to cytotoxic agents to those patients who are likely cured from surgery alone. Recently investigators have tried using molecular markers to try to identify those patients at higher risk: the Her-2/neu oncogene is one such example. At the 26th annual San Antonio Breast Cancer Symposium, several interesting papers were presented that looked into the use of prognostic and predictive factors for breast cancer. One particular paper touted the development of a novel, commercially available assay that can help assess the probability of breast cancer recurrence by ascribing a "Recurrence Score".

Genetic signatures

A novel twenty-one gene, real-time, reverse transcription-polymerase chain reaction (RT-PCR) assay, developed by Genomic Health Inc, preformed favorably in a large scale, prospective multi-center validation study presented at the San Antonio Breast Cancer Symposium. This investigation succeeded in reliably separating patients into two groups: low and high ten-year risk of disease recurrence. Such information should enable the physician to reserve chemotherapy only of those patients who are most likely to derive a benefit – the higher risk patient population.

A big coup for the new assay is that it is able to use paraffin embedded, or fixed, as opposed to fresh, tissue samples to run this analysis. Tissue handling is a very important point, as some preservation techniques are not ubiquitously available. Paraffin preservation is a technique used in all pathology labs for evaluating the breast tumor under a microscope. By contrast, gene microarray assays, although they show tremendous potential in predicting disease recurrence, require special tissue handling that is impractical outside a large tertiary care institution.

This novel assay, known as OncotypeDx, is likely to become available for routine use sometime this month. It will be preformed by a central clinical reference laboratory in Redwood City, California.

Inventors of this novel assay were able to use the National Surgical Adjuvant Breast and Bowel Project?s (NSABP) extensive patient database to design a prospective, blinded, multi-center validation trial. The NSABP is a non-profit clinical trials cooperative group, which includes a network of many health care professionals located throughout the US and Canada. Research conducted by the NSABP is supported primarily by grants from the National Cancer Institute (NCI). www.nsabp.pitt.edu

The following was paper presented at San Antonio:

Multi-gene RT-PCR assay for predicting recurrence in node negative breast cancer patients – NSABP studies B-20 and B-14

Presenter: Paik S

Development for this assay was largely driven by the fact that currently available treatment options for lymph node negative, estrogen receptor (ER) positive patients are based on predictive factors that have limited predictive power or are not reproducible from study to study. There is a clear need for better and more robust markers that would ideally be able to stratify all patients into either a high or a low relapse risk group. Therapy could then be tailored to each of these groups of patients. The OncotypeDx development and validation study was a multi-step effort. First, the assay itself, quantifying gene expression was developed. This assay is unique in that it can simultaneously examine multiple genes for overexpression using paraffin embedded tissue blocks as opposed to fresh samples. Once the technique was perfected, representative genes were selected from the human genome using microarray technology. Investigators note that twenty-one genes were selected from 250 candidate cancer related genes. Using data from three different previously reported clinical trials; the study investigators assured themselves that the chosen genes were indeed overexpressed in relapses breast cancer. Finally, using tissue samples from another large NSABP clinical trial, investigators prospectively validated the "Recurrence Score" as a reliable predictive measure of disease.

Assay development

The OncotypeDx assay calls for ribonucleic acid (RNA) extraction and purification from tissue specimens. Next, the RNA is analyzed using a technique called RT-PCR. RT-PCR enables scientists to analyze gene expression of several hundred genes at the same time. It is extremely sensitive, specific, and highly reproducible. OncotypeDx, however, analyzes overexpression of twenty-one selected genes in the patient RNA sample. Using an algorithm that takes into account the expression of the twenty-one genes in the tumor sample, each patient is assigned a "Recurrence Score" on a scale from 0 to 100. The Recurrence Score correlates to the likelihood of disease relapse. Additionally, the Recurrence Score categorizes patients into a risk group – low, medium, or high risk. A score of <18 is considered to be in the low risk category. A recurrence score equal to or over 31 puts the patient in the high-risk group.

Gene choices

Investigators examined an initial pool of over 40,000 candidate genes. The list initially narrowed to 250 genes using cancer literature, microarray data, genomic databases and molecular biology information. These 250 genes were tested in 447 patients for validation and further reduction.

Validation of choices

The next step was to test the 250 candidate genes in patients. Three breast cancer studies were used to assess the reliability of these genes in predicting recurrence. The largest of these three trials was the NSABP B20 trial that included 233 node negative, ER positive patients, all of whom took Tamoxifen for five years. The other two studies included ER negative and lymph node positive patients (Rush Presbyterian, Chicago IL. study and Providence St. Josephs Hospital, Burbank, CA trial). Available tumor blocks on these patients were examined. RNA was extracted and gene expression was evaluated. A univariate analysis of the genes found 41 genes associated with relapse free survival. Some of these genes included Grb-7 (p<0.0001) and survivin (p= 0.0002). In the end, twenty-one genes were selected as the most robust predictors for relapse: 16 cancer genes and five reference genes. Please see below for a list of the genes.

Prospective validation

A large prospective validation trial was conducted using data from the NSABP B-14 trial. The aim of this study was to validate the Recurrence Score as a predictor of distant recurrence in node negative, ER positive, Tamoxifen treated patients. This study looked only at those patients who were randomized to receive Tamoxifen in the NSABP B-14 trial. Patient tissue samples were probed by RT-PCR for the pre-specified 21 genes. The individuals handling the tissue specimens were blinded to patient outcomes. Unfortunately, due to lack of tissue availability, out of the 2,500 participants in the NSABP B-14 trial, only 668 patients were evaluable in this validation effort. Investigators were quick to point out that despite the low number of people, this subset of patients did accurately represent the total population in the NSABP B-14 trial by the major characteristic examined like tumor size and patient age.

At a mean follow-up of 10.9 years, the Recurrence Score accurately predicted risk of disease recurrence. Investigators showed that ten-year disease free survival in those patients with a low recurrence score was significantly better than the ten-year disease free survival in the patients with a high recurrence score. The ten-year rate of recurrence was found to be 6.8% in the low risk group (Recurrence Score <18) and 30.5% in the high-risk group. (Recurrence Score >31). These results were found to be highly significant (p<0.00001). Using statistical calculations, investigators were further able to prove that a Recurrence Score is a more powerful predictor of distant recurrence than primary tumor size or patient age (p<0.00001). It is also a more powerful predictor than tumor grade, as the latter is subjective and varied between different readers. Recurrence Score, although correlated modestly with tumor grade, is an objective and reproducible prognostic factor that is much more powerful than tumor grade.

In summary, the NSABP B-14 trial has validated the Recurrence Score in ER positive, lymph node negative, Tamoxifen treated patients. Recurrence scores are thought to provide accuracy and precision in predicting the likelihood of distant disease recurrence. In addition, Recurrence Scores are found to be stronger prognostic tools in assessing risk of relapse than standard measures such as patient age, tumor size, and tumor grade.


There are, however, a few limitations to the OncotypeDx. First off, the majority of the twenty-one genes used in the RT-PCR screen are those associated with either estrogen and cell proliferation They represent only additional prognostic markers and do not indicate which patient may benefit from more aggressive upfront therapy. Second, the significant differences seen in this analysis are only between the low and high-risk groups. Therapy decisions for patients with an intermediate Recurrence Score should be based on standard predictive factors used previously. Finally, it is crucial to recognize that this assay applies only to ER positive, lymph node negative patients treated with Tamoxifen. Thus a new patient who is found to be lymph node negative and ER positive could send her tissue specimen for evaluation and the Recurrence Score would only be applicable if she would take Tamoxifen for five years. Put another way, the score cannot influence the decision of whether or not to take Tamoxifen.

Perhaps the biggest criticism of the study is that the prospective validation analysis was carried out in an inappropriate group of patients if the intended question is to evaluate the benefit of chemotherapy in the high Recurrence Score group. None of the participants in the NSABP B-14 trial received any chemotherapy as part of the trial. Therefore, a high Recurrence Score still does not immediately translate to a benefit from system chemotherapy.

Genomic Health Inc.

Genomic Health Inc, will make this test available sometime this month or next. It has not yet received FDA approval and the cost is yet to be determined.

The 21 selected genes: Best RT-PCR performances and most robust predictors.

  1. Proliferation genes
    1. Ki-67
    2. STK15
    3. Survivin
    4. Cyclin B1
    5. MYBL2
  2. Invasion genes
    1. Stromolysin 3
    2. Cathepsin L2
  3. Her-2/neu related genes
    1. Grb7
    2. Her-2
  4. Estrogen receptor related genes
    1. ER
    2. Progesterone receptor
    3. BCL2
    4. SCUBE2
  5. MIS genes
    1. GSTM1
    2. CD68
    3. BAG1
  6. Reference Genes
    1. Beta-actin
    2. Gapdh
    3. RPLPO
    4. GUS
    5. TFRC