Breast Cancer Update for Surgeons, Issue 1, 2017 (Video Program)Critical evaluation of the MINDACT trial results
8:17 minutes.
TRANSCRIPTION:
DR SPARANO: The schema of the MINDACT trial, it’s a complex trial that was designed to try and address a number of questions. But it was a large study. It included about 6,700 patients. All of the patients had a clinical and genomic classification performed. The clinical classification was by the Adjuvant! Online model. And a low risk, as projected by that model, of 92% or lower was considered a clinical low risk. And they also underwent the 70-gene assay, the MammaPrint assay, which is a binary test and provides a result of high or low risk. So what they did was, they looked at the proportion of patients where both the clinical and genomic risk was low. So 41% of the patient population had both a clinical and low genomic score. And about 27% of patients were high risk by both clinical and genomic criteria. The real focus of the study were the patients where there was discordance between the clinical risk and the genomic risk. Nine percent of the patients were clinically low risk but genomically high risk. Twenty-three percent of patients were clinically high risk and genomically low risk. So if one just looks at this and makes a decision in terms of recommending chemotherapy, one would recommend chemotherapy to about 50% of patients here, just based on the clinical factors. If one used the genomic factors alone, then that will be about 36% of patients would be recommended. So the take-home message here is that application of this test could result in a 14% absolute reduction in the need for chemotherapy. The primary endpoint of the study was the group of patients who had discordance that were clinically high risk, but genomically low risk and who were randomized to receive no chemotherapy. DR LOVE: Could I just say, just also, I guess we should keep in mind that unlike the TAILORx study and the initial data from the Oncotype, this included patients with node-positive disease. So when they call it high risk, a lot of those were because they were node-positive. DR SPARANO: Exactly. In fact, 48% of the patients in this group had node-positive disease. And that’s what often drives the Adjuvant! Online model in terms of assigning a high risk. So in a sense, this is similar to the TAILORx low-risk registry, in that it’s a group of patients who were identified and assigned a therapy — here, no chemotherapy, just endocrine therapy — and were followed. And so it turns out that this was about 10% of the entire study population. And this was the primary endpoint of the trial. In terms of the timing of when this analysis would be done, it was prespecified that this analysis, this primary analysis, would be done when the 2 conditions were met, when the standard error for distant metastasis-free survival at 5 years met a certain threshold and when at least a third of the patients had been followed for at least 5 years. It turns out that when they reported it, about two thirds of the patients had been followed for at least 5 years. And what this showed was that the recurrence rate, distant metastasis-free survival in this group was about 95%, which was statistically significantly lower than the 92% that they projected. So they did meet the primary study endpoint. So this slide shows the characteristics of the patient population who were randomized, which was 48% node-positive. Fifty-eight percent had a tumor more than 2 centimeters. About 30% had Grade 3 tumors. The median age of the population was about 55. And this slide also shows what the compliance was in the 2 arms. Now again, this is not actually the group of patients in the primary endpoint. This is the group of patients who were randomized. But nonetheless, the primary statistical endpoint was met. They also evaluated the outcomes in the overall group of 1,500 patients who were randomized to treatment by clinical or genomic criteria. And if one looks at the chemo versus no-chemotherapy group for distant metastasis-free survival, there were 22 events in the chemotherapy arm, 37 in the no-chemotherapy arm. That translated into distant metastasis-free survival rates of about 95% versus about 97%. The hazard ratio was 0.65, so it wasn’t statistically significant, but it does suggest that there might be some benefit from chemotherapy here. DR LOVE: And I guess that point was brought up by Cliff Hudis in the editorial in the New England Journal there. It is, quote, not statistically significant, but you look at those numbers, there’s a third reduction in hazard rate, and theoretically it’s a percent and a half absolute benefit. Any thoughts about that? DR SPARANO: I think this trial does provide reassuring information that a multiparameter gene expression assay like the MammaPrint can provide additional prognostic information. In general, these assays are driven by proliferation genes and estrogen receptor signaling genes. So they’re driven by the same genes. On the other hand, they don’t uniformly classify patients the same way. So it really comes down to 2 issues. One is, which assay more accurately classifies the low-risk patients who have a very low likelihood of benefiting from chemotherapy? And secondly, is the trial accurately designed, at least the randomized component of the trial, to adequately exclude a benefit from chemotherapy? But at a higher level, in terms of the concept, the principle that the trial was testing is, is classification by a genomic criteria or gene expression criteria better than clinical criteria? And I don’t really think that’s how these assays should be used, because clinical criteria are actually pretty good, but they’re not good enough. So we use these assays in a complementary way, not in a way that would replace clinical features. I don’t think these assays were developed with the intent or will ever replace clinical features. Their real intent is to be used in a way that complements and adds information to clinical features. |