First comparison of common breast cancer tests finds varied accuracy of predictions

Commercially-available prognostic breast cancer tests show significant variation in their abilities to predict disease recurrence, according to a study led by Queen Mary University of London of nearly 800 postmenopausal women.

The analysis compares four common prognostic tests for breast cancer and finds that not all are able to accurately predict whether cancer will return after five years. This could be limiting clinicians’ abilities to advise patients on whether chemotherapy or preventive medication, in the form of endocrine therapy1, should be taken.

Lead author Dr Ivana Sestak from Queen Mary’s Wolfson Institute of Preventive Medicine said: “This is the first time that anyone has directly compared the prognostic performance of these four common commercially-available tests. This gives clinicians and oncologists the opportunity to review all the results and decide upon the test they want to use for their breast cancer patients.

“Being able to accurately predict the risk of breast cancer recurrence is even more important now that we are in an era where women are prescribed preventive endocrine therapy for many years. If we were better able to accurately assess a woman’s long-term risk, then some women may be able to end their endocrine therapy after five years. But for those deemed high risk of a late recurrence, continuing their endocrine therapy would be a valuable option.”

The team directly compared four prognostic tests for breast cancer in the same patient cohort of 774 postmenopausal women with the most common type of breast cancer – oestrogen-receptor positive (ER+), HER2-negative. They did this by compiling test results from the original tumour biopsies and then looked at whether the patients’ cancers later returned.

The tests compared were Oncotype Dx Recurrence Score, PAM50-based Prosigna Risk of Recurrence Score (ROR), Breast Cancer Index (BCI) and EndoPredict (EPclin), which all work by looking at the expression levels of multiple genes related to breast cancer to predict risk of recurrence.

All major practice guidelines endorse some of these prognostic tests to aid risk prediction in a range of breast cancer patients. In the UK, current National Institute for Health and Care Excellence (NICE) guidelines do not recommend any of the above tests apart from Oncotype, but last month it announced a reversal on its Oncotype recommendation which is currently going through consultation.

Out of the four tests, the US Food & Drug Administration has approved ROR and the European Society for Medical Oncology recommends the use of Oncotype, ROR and EPclin.

The team found that all of the four commercially available tests are providing substantial prognostic value for women with node-negative disease2, by being able to predict the recurrence of breast cancer up to 10 years after the start of their treatment (with ROR, BCI, and EPclin performing better than the older Oncotype test).

However, one important critical time period to consider is years 5-10 after the start of treatment, since over 50 per cent of women with ER+ breast cancer develop a recurrence after five years. BCI, ROR, and EPclin were found to be the most accurate for stratifying women into low or high risk of recurrence after five years in patients with node-negative disease.

For women with node-positive disease2 only those tests (ROR and EPclin) that incorporated clinical information (e.g. tumour size and the number of positive lymph nodes a woman has), as well as gene expression levels, provided substantial prognostic value.

The study is limited in that the results are only applicable for chemotherapy-free and postmenopausal women.

The study has been published in JAMA Oncology.