The EVOFLUx model accurately captures fCpG data patterns. Credit: Nature (2025). DOI: 10.1038/s41586-025-09374-4

An international team led by the Clínic-IDIBAPS-UB along with the Institute of Cancer Research, London, has developed a new method based on DNA methylation to decipher the origin and evolution of cancer, which makes it possible to predict its future clinical course. The study, published today in the journal Nature, analyzes the evolution of the tumors of 2, 000 patients with leukemias and lymphomas.

The study was coordinated by Iñaki Martín-Subero, adjunct professor at the UB's Faculty of Medicine and Health Sciences, ICREA researcher and head of the Biomedical Epigenomics group at IDIBAPS, and member of the CIBER Area for Oncology (CIBERONC), along with Trevor Graham, director of the Center for Evolution and Cancer at the Institute of Cancer Research, London. The first authors are Calum Gabbutt and Martí Duran-Ferrer, and other researchers from Spain and the United Kingdom as well as Sweden, Switzerland and the United States have collaborated.

Reconstructing the evolutionary history of cancer through epigenetics

Cancer does not start at the time of diagnosis, but often develops silently for years. Similar to a plane's black box, which records flight data such as origin, direction and speed, researchers have discovered that the evolutionary trajectory of cancer—metaphorically called "the black box of cancer"—is encoded in the epigenome. In particular, it is registered in a special type of epigenetic mark known as fluctuating methylation.

Although methylation has traditionally been considered to act as a switch that turns gene expression on or off, this study reveals an additional function of this modification. The research team has discovered that the original cell that gave rise to the tumor leaves a unique methylation signature, an imprint that not only reveals the identity of the tumor cells, but also changes as the tumor grows and diversifies.

Using advanced mathematical modeling, the team deciphered these methylation patterns, reconstructing both the origin and the evolution of the tumor with unprecedented accuracy, which also allows predicting the future progression of the disease.

EVOFLUx reveals the evolutionary dynamics of lymphoid cancers. Credit: Nature (2025). DOI: 10.1038/s41586-025-09374-4

The developed algorithm, called EVOFLUx, was applied to 2, 000 samples from patients with different types of leukemias and lymphomas. "We analyzed old epigenetic data from a completely new perspective, " says Calum Gabbutt, from the Institute of Cancer Research, London. "What we used to consider background noise now revealed the evolutionary history of cancer, " adds Martí Duran-Ferrer, from the UB Faculty of Medicine and Health Sciences and IDIBAPS.

"This new tool allows us to read the past history of the cancer and to know when the tumor originated, how fast it has been growing and whether the tumor has created cell diversity. This is important not only for better understanding the biology of cancer, but also has clinical applications, " adds Martín-Subero.

Predicting the clinical course of cancer years in advance

Based on the hypothesis that knowing the past of a cancer can anticipate the clinical future, the study analyzed samples from patients with lymphoid cancers, including pediatric leukemias such as acute lymphoblastic leukemia and adult diseases such as chronic lymphoblastic leukemia.

Thanks to access to anonymized medical records, the researchers correlated the past evolution of the tumor with its aggressiveness. "Cancers change over time, which complicates their treatment, " says Graham. "We discovered that the initial growth of the cancer determines how it will evolve in the future, allowing us to predict how the disease will progress in each patient. This is a big step in personalized disease management."

Martín-Subero adds, "In the case of chronic lymphocytic leukemia, a type of cancer that does not always require immediate treatment, with this new test we can predict when the disease will need to be treated years in advance, " and he adds that "although in this study we analyzed leukemia and lymphoma samples, we believe that this methodology could work with all types of cancer."

More information: Calum Gabbutt et al, Fluctuating DNA methylation tracks cancer evolution at clinical scale, Nature (2025). DOI: 10.1038/s41586-025-09374-4  Journal information: Nature