Translating sequencing-based tumour profiling and prediction models to the clinic

DNA and RNA sequencing are now standard tools in cancer research but are not yet widely implemented in the clinical setting. Translation of sequencing-based molecular phenotyping to the clinic holds the promise to provide a quantitative and more detailed basis for diagnosis and prognosis compared to conventional pathology, and thus lead to improved patient outcomes.

In this talk I will present results from the ClinSeq (Clinical Sequencing of Cancer in Sweden) breast cancer study. ClinSeq aims to develop and evaluate infrastructure, bioinformatic methods and statistical models to bring sequencing-based cancer diagnostics to the clinic. Using multivariate prediction modelling we assess to what extent DNA and RNA sequencing-based molecular profiling of primary breast cancer tumours can replace, and potentially augment, current routine pathology, including immunohistochemical markers and histological grade. Our results, based on a retrospective study including DNA and RNA sequencing of ~300 primary breast cancer tumours, indicate a high degree of concordance between conventional markers and predictions from models based on sequencing data, while also providing additional treatment relevant information. Around 8% of cases are reclassified by sequencing-based models and these cases also have an increased probability of reclassification under pathological re-examination.