Using SCID-Hu mouse models, Dick and colleagues showed that only 1/250 000 AML CD34+CD38– cells were capable of establishing leukaemic haematopoiesis in the recipient [21,22]. These cells could be targeted by alloreactive T cells recognizing minor antigens on the leukaemia stem Sorafenib nmr cells [7,8]. These models should be interpreted with caution, as the

xenogeneic milieu of the recipient mouse underestimates the number of cells capable of self-renewal and do not provide clear evidence that long-lived AML progenitors are subject to the same degree of immune attack. Furthermore, they do not identify whether all subtypes of AML have comparable hierarchies of long-lived progenitors. Indeed, an alternative model of leukaemia cure is that a sustained T cell response to the progeny of the AML stem cell but not the small stem cell pool itself could contain the leukaemia at a minimal disease level, resulting in a functional cure [3]. Although the concept of immune surveillance is well accepted, evidence for IS specifically in AML is largely indirect, revealed through relationships between treatment outcome and learn more immune parameters and adaptive changes made by the leukaemia favouring immune evasion, unlike viral-induced malignancies. Perhaps the most compelling evidence for a significant role of immune control of AML comes from several observations indicating that

lymphocyte recovery following induction chemotherapy is strongly predictive for outcome. T cells are reduced after chemotherapy mafosfamide but have a rapid clonogenic potential which allows a swift T cell recovery [23]. Patients achieving the highest lymphocyte counts within 6 weeks of chemotherapy have the lowest relapse rates [24–26]. Long-term survival in AML is also favoured by normalized lymphocyte counts [27]. These data all suggest that an intact immune system can protect against relapse of disease, but do not define whether the effect is mediated through T cells or NK cells. There are diverse abnormalities in AML at presentation and relapse that suggest how the leukaemia may develop despite immunosurveillance and how an established leukaemia may acquire new characteristics to defeat immune control. Figure 1 depicts the interactions between AML cells and the immune environment. Genetic features are emerging that may favour the development of AML in the presence of an intact immune system. There is an increased frequency in AML of a particular genotype of the co-stimulatory molecule cytotoxic lymphocyte antigen -4 (CTLA-4) [28]. The inhibitory KIR molecule KIR 2DL2 is expressed more frequently in AML, again suggesting a predisposition for AML through some form of immune escape [29]. There is also strong evidence that an established AML can mutate to escape immune control.