So, one of the interesting things about the community where we reside, the scientific community, is that the TAILORx study, when it was first published many, many, many years ago, it didn’t include histology. So we had no clue about these different histological phenotypes, whether they behave differently when they have the same genomic risk score. Now, what does this mean? We know that breast cancer is a family of diseases...
So, one of the interesting things about the community where we reside, the scientific community, is that the TAILORx study, when it was first published many, many, many years ago, it didn’t include histology. So we had no clue about these different histological phenotypes, whether they behave differently when they have the same genomic risk score. Now, what does this mean? We know that breast cancer is a family of diseases. But the two most important families within that big family are the ductal and the lobular phenotype. So what we did with the TILS working group, and this was a study that was coordinated and led by Joe Sparano and the PI of TAILORx, we have analyzed the 10,000 slides from all those patients to identify ductal versus lobular histology. And we did find that histology, as defined by the pathologist, is a very important prognostic variable. That means that if you are low genomic risk and you are ductal or lobular, if you have a ductal phenotype or a lobular phenotype, that those who have a lobular phenotype have a very bad outcome. They have, at 15 years follow-up, up to 5% lower overall survival compared to the doctor. And this is clinically important. Why? Because those patients today get the advice, because they are low genomic risk, we will not give you chemotherapy. We have shown that those patients with lobular phenotype, that they may actually benefit with more treatment because they are low genomic risk, but may be clinically high risk. In addition to that, to make this diagnosis more precise, we have to combine the pathologist, just like in Affinity, with an AI essay that measures a specific genomic alteration, typically for lobular. And what have we shown is that in the very large majority of cases, pathologists are quite good at making diagnosis of lobular, but there were discordances, meaning it was not perfect, and AI could help us to reduce more this uncertainty of the histological diagnosis. So the combination of the pathologist with AI was actually a very important prognostic variable and reassuring the pathologist that what they do already is qualitatively. And in addition to that, and that’s a second element of AI within the TAILORx, a bit similar like affinity, we did find that patients who had low genomic risk that had this different histology, that if you add a metric of the immune system that we co-developed with case 45, that we could even then identify more patients who were clinically high risk of recurrence so combining histology by the pathologist plus ai plus a metric of the immune system with ai we were able to identify patients who currently today get the advice not to get chemotherapy, but that we believe based on this evidence, thanks to AI, that those patients could benefit from more extensive treatment because they may be clinically high risk.
