Per Karlsson, MD, PhD: New Data on Breast-Conserving Surgery, With or Without Radiotherapy

We have performed a patient level meta-analysis and we have included three different studies, studies randomizing patients to radiotherapy or no radiotherapy. The studies included were the Swede BCD 921 study, the Princess Margaret study, and the Scottish Conservation RT Trial. The patient level meta-analysis has the objective to analyze the performance of the POLAR signature as a prognosticator for locoregional recurrences in the new RT arm and also as a potential predictive assay for radiotherapy benefit. It's well known that radiotherapy reduces the risk for locoregional recurrences in all different clinical subgroups today. There are definitely some need to find new classifiers to understand which patient we can omit the radiotherapy for. We have within the Swedish trial identified a signature for RT remission, but that signature must be validated in further studies. Now we have performed this patient-level meta-analysis, but to send some word about the development of the signature in the Swedish study, we have collected all the blocks from the tumors and we have punched the tissues and isolated RNA, and this RNA has given us gene expression info. We have used array-based methods for that. Using this information, we have filtered the genes most prognostic for locoregional recurrences in the new RT arm. We have further [inaudible 00:01:52] the genes at the enrichment analysis. This filtering has resulted in genes that were fed into a regression model. This regression model resulted in the POLAR score. The POLAR score consists of 16 genes. The gene functions here is about immune response and also about proliferation. Now we have used this POLAR score and applied that in this patient meta-analysis, and we could see that POLAR could prognosticate the locoregional recurrence in all the new RT arm in all these three different studies. We also find that the POLAR RT had a significant interaction, which means that this classifier is a predictive for radiotherapy benefit. But all this has been done retrospectively. So going further to use it in clinical practice, I think we need further validation, but this is, to our knowledge, the first genomic classifier that can predict radiotherapy benefit. So hopefully in the future we can use this assay to know which patient we can omit radiation. So that's the way forward for us.