Gene Expression Technologies Show Promise As An Aid To Breast Cancer Decision Making

Gene expression technologies show "considerable potential for improving prognostic and therapeutic prediction" in early-stage breast cancer concludes a systematic review, published in the Annals of Internal Medicine, of the three currently available commercial assays. "But more needs to be learned about the extent of that improvement..... and how they are best incorporated into decision making about current breast cancer treatment," write the study's authors Steven N. Goodman and colleagues from Johns Hopkins University ( Baltimore, Maryland, USA).

Many women with early-stage breast cancer need to make the decision of whether to be treated with systemic chemotherapy after surgery and hormone treatment. Although adjuvant chemotherapy is frequently recommended, the reality is that many women (especially those with small, estrogen receptor-positive tumours) would have remained recurrence-free at 10 years without it, and could have avoided toxic side effects. Traditionally oncologists base their therapy decisions on factors such as tumour stage, and tumour characteristics, such as grade and estrogen receptor expression. More recently, gene expression profiling technologies have been developed to predict the women whose outcome would be just as good without systemic chemotherapy. Profiling technologies look at the differential expression of genes in breast cancer and how these vary between normal and diseased patients, a! nd women with different prognoses.

In the current review, Goodman and colleagues identified 26 studies, undertaken between 1990 and 2007, involving the three different profiling assays available in clinical practice. Of these, twelve involved the Oncotype DX system, seven the MammaPrint system and six the H/I system. The studies, which were identified from searching the MEDLINE, EMBASE and Cochrane databases, assessed evidence on the prognostic accuracy and ability to predict treatment benefits for each system.

The authors outline fundamental differences between the three systems. The Oncotype DX system, produced by Genomic Health ( Redwood City, California), is based on a 21-gene profile. The MammaPrint, produced by Agendia BV (Amsterdam, the Netherlands), is based on a 70-gene prognostic signature derived from an initially unselected set of more than 25,000 genes. Finally the H/I assay, produced by H/I (AvariaDX, Carlsbad, California) is based on a ratio between the HOXB1 gene and the IL17B gene and uses RT-PCR. The two genes were initially identified from screening 22 000 genes in 60 patients with estrogen receptor-positive, lymph node positive or negative breast cancer being treated with tamoxifen. Results showed high expression of HOXB13 predicted recurrence, while high expression of IL17BR predicted non-recurrence, leading to the development of a predictive ratio. The Oncotype DX and H/I systems utilize the molecular biology technique real-time reverse transcriptase polymerase chain reactions, to identify genes through their RNA products; while MammaPrint uses DNA microarrays to identify the genes through their DNA products. Finally, Oncotype DX and H/I are carried out on formalin-fixed, paraffin-embedded tumour tissues, whereas fresh, unfixed tumour tissue is needed for MammaPrint.

The authors point out that the genes utilised in the different assays show little in the way of overlap. For example the 21-gene signature, forming the basis of the Oncotype assay, and the 70-gene expression, forming the basis of the MammaPrint assays, have only one gene in common. Goodman and colleagues make a key distinction between gene expression signatures, describing the actual genes measured in a given test, and the marketed gene expression test that measures the full range of genes. The reality is that use of different reagents, sample acquisition, preparation and transport procedures, can all restrict the genes that are identified. "It is therefore critical to pay close attention to the test description, population, and end points for each study to understand which studies are mutually supportive, which are adding qualitatively different infor! mation, and to whom they apply," caution the authors.

Goodman and colleagues point out that Oncotype DX is the only technology that has actually examined the gene profile's ability to predict chemotherapy benefit. The study, examining 10 year distant recurrence free survival in 651 patients with estrogen receptor-positive, lymph node-negative disease randomly assigned to receive tamoxifen alone or tamoxifenwith chemotherapy, showed overall benefit from chemotherapy was restricted to patients with high recurrence scores. No such studies have been undertaken to address whether the Mammaprintor H/I systems are capable of predicting the clinical benefits of chemotherapy.

Overall, the authors conclude, the new generation of tests offer improved risk stratification over standard predictors, but the tests are at varying stages of development. The Oncotype DX assay provides the strongest evidence, followed by MammaPrint and finally the H/I test, which they say, "is still maturing". "For all tests, the relationship of predicted-observed risk in different populations needs further study," write the authors.

Future issues that need addressing are the fact that gene expression assay evidence relates to patients treated with tamoxifen, and not current treatment regimens such as aromatase inhibitors, and anti-HER2 drugs. Furthermore, increased use of assays will lead to a need for additional laboratories to undertake tests, which will require the development of systems to monitor their predictive accuracy. Additionally, data bases with complete data on each patient are needed to expand the current pool of information and in it will be important to understand what risk benefit ratios real women find acceptable. Finally, as tests are modified and new ones developed comparative effectiveness studies will be needed to show how they compare with one another and whether combining tests is of value.