The TAILORx study is one of the largest, maybe the largest, study that has been performed to evaluate the prospective validation of the genomic assay to find patients with hormone receptor positive disease, the so-called luminal breast cancer patients, who may not need chemotherapy. Because we do know for many years already that patients who have low genomic risk, identified by a genomic signature, have such a good outcome that they may not need chemotherapy.

Now, two elements in this study. We have analysed in this study whether histology matters and breast cancer is a family of diseases. The two most important histological subtypes are ductal breast cancer and lobular breast cancer. So we have looked at, in this study, these two histologies but not only by the pathologist looking through the microscope but also we have combined an AI assay which is very specific for lobular breast cancer. So we have used manual, a bit like the APHINITY study, a manual assessment combined with an AI assay (Paige.AI), that measures a genomic alteration on the digital H&E slides so that the pathologist who thinks it’s lobular and if the AI tool confirms it is lobular we have nearly 100% assurance that it is indeed what he’s told you, because it’s not always that easy. By doing this we have shown that not the first five years but after five years, ten years, with 15 years follow-up, we see that the overall survival difference for lobular breast cancer is 5% less than for ductal. So this is an extremely important finding because nowadays we tend to think that lobular breast cancer is a breast cancer with an indolent disease, we just have to treat with endocrine therapy. But we do show that if you just wait long enough, those patients live shorter and have a higher risk that the tumour recurs. This may have important implications for those patients.

So nowadays those patients get five years of endocrine therapy, but these findings may be suggesting that a subset of them may need more than five years of treatment and this is a very important clinical finding.

In addition, we do know that some patients with so-called genomic risk who, according to the first results of these trials many years ago, are associated with an excellent outcome so that they do not need chemotherapy. But we have shown, and in clinics it is known, that some patients, even with low genomic risk, recur, so develop metastases very early. We don’t know how to find those. So applying the same AI, as developed by Case45 which we applied in APHINITY, we have found that the immune system is able to find those patients with low genomic risk that nowadays get the advice not to get long treatment, that the immune system is able to find those patients that will recur. This is a very important clinical finding because, again, most of these patients with low genomic risk do not get – again, I repeat, do not get – extended treatment or chemotherapy. We have shown that a subset of them, where the immune system is active, those are the patients that recur and that they may need additional treatment.

So both of these studies, APHINITY and TAILORx, demonstrate the concept that integrating AI to help pathologists and to help clinicians is feasible. It’s very feasible because it brings additional information to them which cannot be provided by the pathologist.