DR RUGO: NSABP-B-28 was a large adjuvant trial in over 3,000 women. Patients were randomized to receive standard doxorubicin and cyclophosphamide with or without paclitaxel given every 3 weeks, as was the standard at that time. As you recall, we had our CALGB trial that presented around the same time period and showed that the addition of paclitaxel improved disease-free and overall survival. The NSABP trial showed an improvement in disease-free survival but not an improvement in overall survival. And subset analyses, I think, then confused the treatment of ER-positive disease for almost a decade subsequent to that, or maybe a whole decade, because it suggested that patients with ER-positive disease, as if they were one group, didn’t benefit from the addition of paclitaxel.
I think subsequently and from additional analyses from the TAC versus FAC trial, we understand that this has to do with the type of ER-positive disease. We have, but nonetheless, as we’ve heard today, it’s become a standard for our adjuvant therapy. And, of course, now we give paclitaxel differently, dose-dense weekly, et cetera, because we believe it to be more effective.
What they did was actually look at tumor blocks that were available and do the 21-gene Recurrence Score. They had, of course, a subset of patients for whom tumor was available, about 1,000 out of 3,000. And then there were some patients who didn’t have successful assays. But I have to say that that was a very small number. There was insufficient RNA in only 11 patients. And the sample quality was poor in 7. And now there’s a median follow-up of 11.2 years. Also, all patients in this trial were required to have node-positive disease.
What they found actually, in this assay, is that the Recurrence Score was very prognostic even in this node-positive population. And it was interesting, actually, to look to see what was the distribution of Recurrence Score among the population as a whole. Thirty percent were high risk. Thirty-six percent were low risk. And about 34% were intermediate risk, which is interesting. It’s about a third, a third, a third. So different from what we see in the node-negative, ER-positive patient population, where almost all the patients have low and intermediate grade, and just a small percentage tend to have a high risk. That was the first thing they learned.
The second thing they learned was that the Recurrence Score in the trial population as a whole, regardless of treatment, was able to predict disease-free and overall survival. The overall survival in low-risk patients was 90%. The breast cancer-specific survival was 95%, compared to an overall survival of 63% and 68% in patients who had high risk. And then the intermediate-risk group fell right in between, so disease-free survival was similar to that, about 76% node-positive disease in low risk versus 48% in high risk.
Then they wanted to say — and this is the elusive and difficult goal — we want to find predictive factors. The question is: Could this predict the benefit of paclitaxel? If you had high-risk disease, would you get more benefit from adding paclitaxel than if you had low-risk disease and, interestingly, didn’t predict benefit of paclitaxel at all. If you looked at patients in terms of the addition of paclitaxel, the groups that appeared to benefit more were in the high-risk group than the low-risk group.
But when you looked at p-values for interaction, they were all nonsignificant. Now, it’s a subset study looking retrospectively, and so it’s possible that the reason why it didn’t show a differential benefit is because of just the fact that there weren’t enough samples and the samples were specific.
We talked about earlier that if you have a higher risk and there’s a certain amount of benefit — let’s say you have a 30% relative reduction — if you have a higher risk, you’re going to get more benefit. And that’s pretty much what they showed here. They showed that regardless of treatment, if you have high-risk disease, a high risk of distant recurrence, that you’re going to have a risk of high-risk of distant recurrence no matter what your treatment is, even though we keep chipping away at it.
DR LOVE: Any other thoughts about the study? Was it adequately powered? It seemed that the entire study was borderline powered, and yet this is just a small sample of it. Was it powered enough that they should have seen a benefit, if it was there?
DR RUGO: I think that there are a number of problems. One is it’s a subset where you have tumor samples available. That’s always going to be an issue. The other is that everybody got concurrent tamoxifen. And that may also affect benefit. I think that my conclusion from this is that the Recurrence Score is prognostic in patients with node-positive disease treated with chemotherapy. The predictive nature of it remains to be seen. I think it’s going to not be a kind of score that’s going to predict relative benefit of one chemo versus another. But it may be that if you have a high score, you benefit from more versus less.