Cancer is a complex disease that, despite decades of research, still poses a major health problem worldwide. Although there is a fundamental understanding of the molecular mechanisms underlying carcinogenesis, the transition from healthy to cancerous tissues is a multifactorial process that remains poorly understood. For example, it is unclear why healthy epithelia carrying oncogenic driver mutations can remain healthy in some instances but not in others1.

The limited understanding of cancer initiation is due largely to the inability to effectively model tumorigenesis in the laboratory. Although the currently available cancer cell models are suitable systems for studying simple cancer cell behaviours, they lack the cellular diversity, tissue-level organization, longevity and experimental versatility needed to capture the intricate process of tumorigenesis. Therefore, researchers have traditionally relied on inducing tumours in animal models as a solution. Even though this approach has provided highly valuable insights into tumour biology, it is inadequate in terms of resolution. Animal models often function as black boxes in which tumours are induced and data are collected after a certain period of time, missing crucial real-time and single-cell resolution information on the tumorigenic process. Furthermore, the use of animals as research tools is associated with ethical considerations.