Improving our ability to predict how a tumour will evolve by overcoming the limitations of analyses that only consider individual genetic mutations. This goal is possible thanks to a new method called ASCETIC (Agony-baSed Cancer EvoluTion InferenCe), developed by Milano-Bicocca, which is capable of reconstructing patterns of tumour evolution for each patient and subsequently identifying evolutionary patterns that are repeated in different patients.
The method, described in the article "Evolutionary signatures of human cancers revealed via genomic analysis of over 35,000 patients" just published in Nature Communications, was developed by a multidisciplinary team led by Daniele Ramazzotti, Professor of Informatics at the Department of Medicine and Surgery of the University of Milano-Bicocca, in collaboration with Alex Graudenzi (Department of Informatics), Luca Mologni (Department of Medicine and Surgery) and the researchers Diletta Fontana, Ilaria Crespiatico and Valentina Crippa, for the evaluation and validation of the results.
In this study, ASCETIC was applied to data from over 35,000 tumours, including patients with various blood disorders, patients with early or advanced lung cancer and many others. In addition, the results were validated against independent datasets to ensure their reliability and generalisability.
Cancer is a complex developmental process involving large populations of cells in the human body. These cells undergo genetic mutations and epigenetic modifications, some of which can give cancer cells an advantage. This advantage can translate into an increased ability of cancer cells to proliferate and survive, which can eventually lead to invasion of surrounding tissues and the formation of metastases. However, not all mutations contribute to the development of the disease. In fact, only a small fraction of them, called 'driver mutations', play a functional role, while most mutations are neutral, called 'passenger mutations'.
ASCETIC is based on the observation that, in most cases, the accumulation of passenger mutations during cancer progression follows a random dynamic. However, for driver mutations, which are responsible for tumour progression, evolution can lead to a consistent order observed in different patients.
ASCETIC addresses this complex problem by breaking it down into three key steps. First, it uses evolutionary models to establish an order between driver gene mutations in individual patients, allowing us to understand the sequence in which these mutations occurred throughout the evolutionary history of specific tumours. It then uses artificial intelligence approaches to identify the most appropriate model to explain all the individual evolutions, giving us a map of how cancer evolves globally for a particular tumour type. Finally, it categorises patients according to their evolution and checks whether these groups have different survival curves.
Thanks to the increasing availability of biological data from genetic sequencing experiments on cancer patients, and advances in data science and artificial intelligence, we are now able to assess the presence of specific evolutionary patterns for different types of cancer. These patterns, which we can call 'evolutionary signatures', represent the preferred pathways for the acquisition of driver mutations, i.e. functional ones, during cancer evolution and may recur in patients with similar prognosis.
"Although this study is not definitive," concludes Ramazzotti, "it represents a significant step towards the creation of a 'catalogue' of evolutionary signatures of cancer, which could help to better understand the complex nature of cancer and improve predictions of its progression and prognosis. Indeed, the ability to classify cancer patients according to their molecular evolution could allow the prediction of future disease progression and, consequently, the implementation of optimal, personalised treatments".