The occurrence of concordant childhood leukemia arising in twins was first recorded in the medical literature in 1882 in Germany [10]. By the middle of the 20th century, several case reports of concordant leukemia in twins had been noted, and based on these findings, review papers published in the 1960s and 1970s proposed the prenatal origins of childhood leukemia [10,11,12]. By the beginning of the 21st century, over 70 twin pairs with concordant childhood leukemia were reported in the literature, including at least 37 ALL and 17 AML twin diagnoses [10]. The majority of the reported concordant leukemias in twins were diagnosed in children younger than 5 years of age, with approximately half of the concordant pairs being diagnosed with infant leukemia (i.e., <1 year of age), a proportion of total childhood leukemias much higher than that found in singleton cases.

The concordance rate for acute leukemia in monozygotic (MZ) twins, reported to be between 5% and 25%, is vastly higher than the concordance rate of leukemia in dizygotic, or fraternal, twins, in whom concordant leukemias are exceedingly rare. Such a large difference was unlikely to be due to a shared genetic predisposition in MZ twins, and instead, a proposed mechanism that could plausibly explain the high concordance rate of leukemia in MZ twins was their shared placental vasculature [10,12,13]. Approximately 60% of MZ twins develop in a single monochorionic placenta, which leads to vascular anastomoses that facilitate the passage of blood cells between the circulation of each twin [14]. Thus, it was proposed that an initiating leukemogenic lesion could develop in one twin and then be transferred to the other through the shared placental vasculature [12].

This hypothesized mechanism of twin-to-twin transfusion of preleukemic clones was subsequently confirmed by molecular genetic analyses of acute leukemias in concordant twin pairs. In particular, the detection of identical fusion events in MZ twins concordant for leukemia supports a common clonal origin, given that the independent generation of leukemia-forming translocations with identical breakpoints in both twins is highly unlikely. Ford and colleagues first demonstrated this in three pairs of MZ twins concordant for KMT2A-rearranged infant ALL, whereby each twin pair shared the same-sized restriction fragments of KMT2A [15]. 

Studies of leukemia in twins have also been informative regarding the timing of mutational events that initiate or drive the progression to overt leukemia in children. In twins concordant for infant leukemia, the latency period between the diagnosis of leukemia in both twins is relatively short. However, a wide variation in age of leukemia diagnosis has been observed in twins concordant for childhood ALL, with latency periods of up to 14 years reported in one twin pair [10,17]. This supports a two-hit mechanism of leukemogenesis, whereby the initiating lesion develops prenatally, and then secondary somatic mutations arise in the postnatal period, resulting in progression to acute leukemia. Genetic studies in twins concordant for ALL have provided strong evidence for this two-hit model, with several reported instances of distinct secondary somatic mutations and copy-number alterations in twins that share the same gene fusion or other initiating events [22,24,25]. Further support is provided by twins discordant for childhood ALL, in which preleukemic clones harboring the initiating lesion, such as ETV6-RUNX1 fusion, have been found to persist for several years in the healthy co-twin without leading to the development of leukemia, supporting the supposition that additional mutations are required for overt leukemia development [19,24,26,27]. The timing of mutational events that drive the development of childhood AML is less clear, as fewer twin pairs with AML have been investigated, although a protracted latency period between the onset of AML in at least one concordant twin pair has been reported [28].

3. Prenatal Risk Factors for Acute Leukemia in Childhood