The new test, called the MammaPrint test, developed in Amsterdam, in the Netherlands, looks at the "gene signature" from 70 cancer-related genes in a breast tumour.

On the basis of the genes seen in the cancer, a woman can be seen to be at high or low risk of having her breast cancer come back after surgery.

If the test says a woman is at a low risk it means the woman has a 95% chance of remaining cancer-free after five years and a 90% chance of remaining cancer-free after 10 years without chemotherapy.

If the test says a woman is at a high risk it means the woman has a 23% chance of having her cancer come back within five years and a 29% chance of having her cancer come back after 10 years and such women are more likely to need chemotherapy.

Research suggests the test could cut the number of women who have to undergo breast cancer chemotherapy by 20% to 30%.

Dr. Steven Gutman, director of the FDA's Office of In Vitro Diagnostic Device Evaluation, says the test provides powerful new insight into the behaviour of the tumour, but the decision whether or not to undergo chemo should be based on other clinical and diagnostic factors in addition to the MammaPrint test.

Gutman says the test is a good test but is not a perfect test.

In order to obtain a MammaPrint test, a doctor must send a piece of the breast tumour removed during surgery to the Amsterdam lab.

Scientists at the lab then use a computerized tool, called a microarray, which tests the tumour for activation of 70 different genes linked to breast cancer; the results are ready in two days.

MammaPrint has been tested on 307 women under age 61 who underwent surgery for stage I or stage II breast cancer that had not spread to the lymph nodes.

That study showed the test more than doubled the doctors' ability to predict a breast cancer recurrence.

A major European clinical trial is currently testing whether the test really helps women avoid needless chemotherapy with 6,000 women recruited who have early-stage breast cancer.

This year alone more than 178,000 women in the U.S. will be diagnosed with breast cancer; more than 100,000 of them will have early stage tumours with a remarkably good prognosis because they are small and have not spread to the lymph nodes.