Influence of Different Genetic Alterations On the Engraftment Capacity and Serial Transplantation Efficiency of AML Patient-Derived Bone Marrow and Peripheral Blood Cells in NSG Mice

In order to allow a better understanding of acute myeloid leukemia (AML) and develop more promising therapeutic strategies the establishment of functional and reproducible in vivo models is widely pursued. Of available model systems, xenografts in immunodeficient mice reproduce the clinical situation best. Here, we performed extensive analysis of AML engraftment in NOD/SCID-IL2-receptor-gamma-chain^−/− (NSG) mice comparing tail vein versus intratibial injection and growth behavior of AML patient-derived bone marrow versus peripheral blood cells. Furthermore, tumor growth characteristics in the murine host were correlated with the disease stage and the molecular risk factor profile of the individual donors.

Bone marrow and peripheral blood cells from 17 AML patients were injected intratibially into NSG mice (n=4–8/patient, 82 mice in total). As controls, 14 mice received bone marrow from three different donors and 5 mice were mock-injected. Tumor growth was monitored via a) determination of overall survival, b) fluorescence-based in vivo imaging (IVI, Kodak FX, Alexa750 labeled anti-human CD45 or CD33 and c) confirmation of IVI data by histological and immunohistochemical examination of bone marrow and spleen. When highly positive IVI signals and/or the overall condition of individual mice indicated enlarged tumor burden, the respective animals were sacrificed and the human AML cells transferred to another animal. In parallel the engraftment pattern of AML cells 2–4 weeks posttransplant was correlated with clinical disease activity, application route and origin of the particular tumor cells.

Patients included in the present study represent multiple different French-American-British (FAB) subtypes, various karyotypes and molecular features in terms of the mutational status of NPM1 and FLT3. All patient-derived specimens were capable of recapitulating the disease in NSG mice at 4–6 weeks after transplantation. Over a period of 13 months 12 out of 17 xenografts could be passaged once and 9 at least twice. Up to six passages were performed for an individual AML xenograft. In contrast, engraftment of healthy donor bone marrow cells could be determined merely until day 56 after implantation. The human bone marrow cells of the healthy donors did not engraft in serial passages. Mean survival time of AML bearing animals ranged between 21 and 82 days for a respective xenograft. No differences could be determined between engraftment capacities of peripheral blood or bone marrow cells of one patient. Neither karyotype, FAB classification nor leucocyte count or the percentage of monomorphic blasts in the bone marrow seemed to have an impact on engraftment capacity in the murine organism. However, mice bearing AML xenografts with mutations in FLT3 as well as in NPM1 showed particular short overall survival times and high tumor cell engraftment determined by IVI. This phenomenon became more obvious along the different passages. The intratibial approach proved to be superior in comparison to the intravenous application as cells of an individual patient engrafted faster when injected directly into the bone marrow microenvironment. Determination or tumor load via IVI permits to closely monitor not only the growth behavior but also the homing characteristics of the human cells over time. The positive IVI signals in bone marrow and spleen could be confirmed by histological examination as well as by immunohistochemistry specific for human CD45 and CD33.

Our xenografts show a close resemblance to the AML-disease regarding the level of dissemination and organ involvement. Collection of whole-body IVI data proved to be a time- and animal-saving analysis that allows to closely monitor AML growth. As AML is characterized by an increasing number of molecular subtypes with completely different therapeutic options it seems to be extremely worthwhile to develop patient derived xenograft models representing as many AML subtypes as possible. Our results suggest that this model reflects the heterogeneity and important clinical characteristics of the disease, and thus may serve as a tool for preclinical drug testing and investigation of the pathophysiology of AML.

No relevant conflicts of interest to declare.