Cancer is a multi-step process, consisting of initiation and transformation phase. During initiation, there is continuous and progressive selection for aggressive clones by acquiring mutations. Apart from mutations in oncogenes and tumor-suppressor, we know surprisingly little regarding the molecular networks that underlie the emergence of cancer-prone cells. This is especially true in adult tumors where initiating mutations have remained largely elusive due to the large milieu of cancer-causing genes present at early stages of the disease. With time, cancer progresses and accrues more mutations, which give the cells an impressive armament to resist chemical and radiation-based therapies. As such, developing effective therapeutics for adult tumors has been challenging and costly. In contrast to adult tumors, paediatric tumors often develop from a single initiating mutation and have predictable outcomes. For instance, newborns that have one inactivating mutation in one RB1 allele (RB1+/-) develop tumors within the first year of life and many of whom die of secondary tumors in the bone, brain, and lung. We can leverage unique cancer properties of retinoblastoma to overcome the challenges of dissecting disease networks, and ultimately, pharmaceutically exploit these networks to prevent heritable cancer. Here, we define molecular properties of cancer-prone cells and use this information to prevent retinoblastoma with only week of targeted therapy (2 months before expected tumor emergence). Surprisingly, the cancer-prone state could be reversed without affecting normal physiological processes. Our current data and future directions suggests that targeting initiation networks may hold great promise towards a cancer-free future in children.