Are genomic profiles refined enough that they should be used routinely to determine which breast cancer patients should receive adjuvant therapy? According to J. Michael Dixon, MD, who will be presenting the contra argument to this question in a debate at the Miami Breast Cancer Conference this week, the answer is: Not yet.
Professor Dixon is a fellow of The Royal College of Surgeons of both Edinburgh, Scotland, and England. He is also an honorary fellow of The Royal College of Physicians of Edinburgh. He is a member of the British Breast Group, the Scottish Cancer Trials Breast Group, the European Association for Cancer Research, is a panel member of the UK National Breast Cancer Coalition, and a representative on the Central European Cooperative Oncology Group.
Dr. Dixon spoke with CancerNetwork from his office in Edinburgh, and shared the following observations on his upcoming debate with Norman Wolmark, MD, chairman of the department of Human Oncology, Allegheny General Hospital, Penn.
“The first thing to say is that there have been a number of genomic profiles that have been produced. Most people, I think, are most familiar with the Oncotype DX, although there are a variety of profiles.
“What you need to look at it is: What's common to all these profiles?
“What's common to all these profiles is that the most important thing they measure is proliferation, the growth of tumor cells. We used to have a prime minister who said that all that mattered in our country is education, education, education. The thing to realize is what matters in genomic profiles is proliferation, proliferation, and proliferation.
“And that's why these genomic profiles are fairly good at identifying a group called 'benefit from chemotherapy.'
“The number one argument is: What these genomic profiles do is that they essentially measure proliferation. There are other ways of measuring proliferation. There is a simple method of just looking at something called T-67. There is a very interesting study that compared two types of chemotherapy. What they found is that the chemotherapy including taxanes was most effective, not surprisingly, in the sub-group of patients who had the highest proliferating tumors.
“So, we can say from other studies, that the reason chemotherapy is affecting some tumors is because these are more proliferative tumors. And that's not surprising that high-risk cancers, as far as genomic profiles are concerned, respond better to chemotherapy and have a better outlook.
“The second argument is: If you look at something like Oncotype DX, what that does is, it measures proliferation and it measures estrogen related genes and its also called a HER2 gene complex. The question is: If you were to measure estrogen receptor, progesterone receptor, HER2, and proliferation and combine all those can you produce results identical to the results you get from the 21-gene array you get in Oncotype DX?
“The answer to that is: Yes, you can.
“A study from the UK compared the German Health Profile with that they could derive from measuring estrogen receptor, progesterone receptor, and HER2, and they could derive exactly the same sub-groups as low, moderate, and high-risk.
“So, what I think what something like Oncotype DX does is that it measures ER, PR, and HER2 and proliferation very, very consistently. One of the impressive things about the 21-gene array is that its very accurate and its all done centrally. But, if you had central measurement of ER, PR, and HER2 and proliferation, you could do exactly the same.
“One of the things we're led to believe is that if you use genomic profiling, you can reduce the number of women who are going to get chemotherapy.
“That was put to the test in a recent study that used a 70-gene array profile, a study called the 'RASTER' study. And what did they find? Surprisingly, the opposite of what you'd expect. They found that not less women would have gotten chemotherapy, but that more women would have gotten chemotherapy. More women would have been prescribed chemotherapy with the information provided by the 70-gene array.
“So, to date, close but no cigar. The reality is that genomic profiles are headed in the right direction, but they don't tell us what we really need to know. Genomic profiles are good at predicting outcome. What we want is profiles that predict response; profiles that predict benefit from each of the individual drugs. As yet, we don't have that. And that's why I'm arguing that, at the present time, genomic profiles to all are not consistent with the data we have, and I don't think measuring genomic profiles for all patients is cost effective.”