2011 SABCS, San Antonio Breast Cancer Symposium, 6-10 December, San Antonio, USA
A test to predict risk of recurrence in DCIS
Professor Joseph Sparano – Albert Einstein Cancer Center, New York, USA
My name is Dr Joseph Sparano, I’m a medical oncologist and Vice-Chair of Oncology at the Montefiore Einstein Center for Cancer Care in New York. What we’ve known for some time about the oncotype DX recurrent score, it’s an approved assay, 21 gene assay, that can help provide both prognostic and predictive information for patients who have oestrogen receptor positive, lymph node negative and now lymph node positive breast cancer. It can provide information as either a continuous variable, higher scores associated with a higher risk of recurrence, or a categorical variable – low, intermediate or high. We also know that patients who have a low recurrent score who have invasive cancer do very well with endocrine therapy, have a low chance of benefitting from chemotherapy and that patients with a high recurrent score derive a great deal of benefit from chemotherapy. We’re uncertain right now about how to counsel patients with a score in the mid-range or intermediate range and a large trial has been completed called the TAILORx trial where patients who had oestrogen receptor positive, lymph node negative, HER2 negative disease, who had a mid-range recurrent score were randomised to chemoendocrine therapy, which would currently be the standard of care in that population, versus endocrine therapy alone. We should have the results of that trial in several years.
Meanwhile, there’s another trial on-going that is targeting patients with one to three positive nodes that’s asking a similar question in that patients who have a lower intermediate score and positive lymph nodes are randomised to chemoendocrine therapy versus endocrine therapy alone. That study is on-going.
What we saw that was new here at the meeting was a prospective validation trial of a DCIS score that uses 12 of the 21 genes in the oncotype DX recurrent score. It was developed by looking at previous datasets and looking at the correlation between gene expression in both in situ and invasive cancer and picking those genes that had a very high correlation between invasive and in situ cancer. Also based on the genes that in other studies had been shown to be prognostic, irrespective of whether tamoxifen was given or not, so purely prognostic genes. Based on those considerations these genes were selected and an algorithm was developed and modelled. This algorithm was then prospectively tested in a group of patients enrolled on a previous trial conducted by the Eastern Co-operative Oncology Group that we call E5194. This study included patients who had ductal carcinoma in situ that was less than 2.5cm in size and associated with a low or intermediate grade, or high grade DCIS that was less than 1cm. These patients all underwent wide local excision with a central review to assure that the margins were at least 3mm and did not receive radiation and were followed. Sometime during the course of the trial it was shown that tamoxifen can reduce the risk of both local recurrence and a contralateral new breast cancer and it was allowed for patients who enrolled in this trial, they were allowed to take it when that information was known. It turned out that about 30% of patients who enrolled on the trial took tamoxifen. It wasn’t randomised so we have really no idea whether or not the event rates would necessarily be lower, the trial was not designed to test that.
The results of this trial were previously presented and published in The Journal of Clinical Oncology with five and seven year results. This analysis represented an update analysis with ten year results and also focussed in on about 50% of the patients who participated in the trial where we had archived tumour material and were able to go back and extract the RNA and performed a prospective validation of this new score. With the cut points, the Cisco plan, the primary endpoint, everything pre-specified. So this is a validation study, it’s a prospective retrospective validation study, that’s acknowledged as an appropriate way to validate a new assay. With regard to the primary endpoint, it was found that the DCIS score, when evaluated as a continuous variable, was significantly associated with an increased risk of recurrence. And when evaluated as a categorical variable also was associated with an increased risk of recurrence in the higher intermediate versus the low-risk groups.
It turns out that about 75% of patients in the study had a low risk, a low DCIS score, and about 25% had an intermediate or high DCIS score. If you look at the recurrence rate in the overall population, in the same breast, it was about 15% and about 10% of that was due to in situ disease, about 5% of it was due to invasive disease. When you then looked at the recurrence rates by DCIS score, there was a significantly higher rate of recurrence in those who had an intermediate or high score versus a low score. On average it was about 25% for intermediate or high versus 12% for the low risk group.
So the take home message, really, about the score is that in patients who you believe would be an appropriate candidate for treating with DCIS, to be treated with local excision alone, that this score can provide you with prognostic information that can separate out those patients within that group who meet those certain clinical criteria, who have a higher risk of recurrence and where treating the patient with surgical excision alone may not be appropriate and where radiation may be a better choice.
On the flipside for patients who meet those same clinical criteria but where the patient or the clinician is leaning towards radiation, this assay can provide information that can identify, again, which patients would have a lower or a higher risk of recurrence and may be useful in making treatment decisions. In the US there are about 45,000 cases of DCIS diagnosed annually, these are women who almost exclusively present with an abnormal mammogram, they have a routine screening mammogram that shows some calcifications, there’s no palpable mass, so they’re asymptomatic. And in the long term, overall they have a very good prognosis with local excision. About 25% are treated with mastectomy in the US, about 75% are treated with a wide excision or lumpectomy. Of those who are treated with a wide excision, about two-thirds receive radiation and about one-third don’t receive radiation. On average, for those who don’t receive radiation the risk of recurrence at about ten years of follow-up is about 30-35%, and for those who do receive radiation it’s about 15%.
The intent of the assay is to provide the clinician and the patient with a tool that can help them make more informed decisions about whether to choose radiation or not. The ideal trial to validate this would be a randomised trial where patients have been randomised to excision plus radiation versus irradiation alone. We don’t have that information right now, patients and clinicians are choosing, are making decisions without this information. About a third of the time they’re coming down on the side of no radiation, about two-thirds of the time they’re coming down on the side of radiation. Oftentimes for those who do get radiated it’s because of other very high risk features. We really don’t have information about the assay in those patients who were not studied in the study, who were not evaluated in the study. So, for example, patients who had tumours more than 1cm and that were high grade, so we really don’t know about the validity of the assay in that population but we have validated the results prospectively in this particular study.
The assay should be available by the end of this month, by the end of December.
