Optimizing Establishment of High Need Pediatric Acute Myeloid Leukemia Patient Derived Xenograft Models

Introduction: Pediatric acute myeloid leukemia (pAML) patient-derived xenograft (PDX) models are vital for pre-clinical evaluation of new agents. New strains of immunodeficient mice (IDM), including NSGS, NRGS, NBSGW, MISTRG/MISTRG6, may improve the low engraftment rates of human myeloid cells. In collaboration with the Children's Oncology Group (COG), we have generated the largest known collection of serially passaging pAML PDX models in the US using these IDM strains (currently 38 pAML models, https://pdxportal.research.bcm.edu). Despite advancements, most established pAML PDX models represent just a few common molecular subtypes, and we still lack PDXs for many high-risk pAML subtypes. Here, we evaluate the efficacy of newer IDM strains of mice and preconditioning with busulfan to optimize engraftment. We aim to characterize engraftment rates across various strains and experimental conditions as well as establish additional serially passaging high-need pAML PDX models.

Methods: Forty patient samples from 11 rare genomic subtypes of pAML, provided by COG, were selected for their high-need genetic fusions (CBFA2T3::GLIS2, DEK::NUP214, ETV6::MNX1, FUS::ERG, KAT6A::CREBBP, MECOM, NUP98::KDM5A, NUP98::NSD1, NUP98::HOXA9, RBM15::MKL1, RUNX1::CBFA2T2(3)). These genetic fusions were selected based on possible availability of targeted agents, lack of available PDX models, and/or dismal current event free survival. Patients contributing samples were enrolled on AAML0531 or AAML1031 trials. One day before sample thaw, NRGS and MISTRG mice are conditioned with intraperitoneal busulfan (40 mg/kg). After thaw, samples are analyzed for viability and patient-specific flow markers (hCD33-PE, hCD45-APC-Cy7, hCD117-BV421). Conditioned mice and 1-3 NSGS, MISTRG, NRGS, and NBSGW unconditioned mice are then tail vein injected (TVI) with 2x10⁵ cells/mouse. Mice are monitored for clinical appearance daily. At 4 weeks post TVI, engraftment is tracked via flow cytometry of peripheral blood (PB) every 2 weeks. Upon reaching 40% PB engraftment or moribund status, PB, bone marrow (BM), and spleen (SP) are harvested and analyzed to measure disease burden, and TVI is done with secondary recipients. Mice that do not reach the engraftment threshold by 26 weeks following TVI have tissues harvested to assess for occult engraftment. AML cells expanded in PDX models are stored for future use and genomic validation.

Results: To date, 11 of 40 vials have been thawed, with 8 vials demonstrating viability in the 59-92% range, suitable for use. One (genomic subtype CBFA2T3::GLIS2) has engrafted. Due to low cell count at thaw (1.67x10⁶), only one mouse/group was injected. An unconditioned NSGS mouse had 3% PB engraftment at 42 days and was clinically ill (hunched posture) at 48 days. Euthanasia and tissue harvest of unconditioned MISTRG and NRGS mice, and a conditioned NRGS mouse harboring this sample showed some engraftment across all 4 mice.

All flow markers (hCD33, hCD45, hCD117) had relatively consistent positive populations within BM, SP, and PB for each mouse. BM disease burden was consistently higher than PB and SP across all mice. hCD117+ was best at discriminating blast% across all experimental mice and is used for subsequent disease burden comparisons. Among the 3 IDM mouse strains tested, the MISTRG mouse demonstrated the lowest disease burden across all tissues (BM: 23.86%, SP: 23.86%, PB: 6.24%). The NSGS mouse had the highest disease burden for SP (80.65%). The conditioned NRGS mouse had the highest disease burden for BM (93.16%) and PB (32.64%). At harvest, SP weight (and cell yield) were 600mg for NSGS (1x 10⁸ cells), 490mg for conditioned NRGS (4.9x10⁷ cells), 230mg for NRGS (1.7x10⁷ cells), and 20mg for MISTRG (1x 10⁵ cells).

Conclusion: This study investigates the potential of new IDM strains and busulfan-conditioning for establishing high-need pAML PDX models. Preliminary data show that our first engrafted CBFA2T3::GLIS2 patient sample engrafted across different IDM strains and without conditioning, but NSGS was optimal for initial engraftment and cell yield. Whether busulfan-conditioning specific IDM strains enhances engraftment rates in certain genetic subtypes of pAML remains to be seen. Identifying optimal engraftment factors for high-need pAML genetic subtypes will enhance PDX model generation and expedite future novel agent pre-clinical testing.

Stevens:Gilead Pharmaceuticals: Research Funding; AbbVie Pharmaceuticals: Research Funding.