Modelling Myelofibrosis in Vivo: Success With NSG-S

Despite substantial progress in our understanding of the genetic evolution of MPNs and in symptomatic management, medical therapies that target the disease-initiating cells and/or prevent fibrosis are lacking. A major obstacle to drug discovery has been the lack of pre-clinical models that adequately recapitulate key features of the human disease, particularly marrow fibrosis. Also, the substantial heterogeneity between patients in terms of risk and rates of disease progression means that identifying patients who may benefit from intensive treatments such as stem cell transplantation or experimental therapies remains a challenge.

Drug discovery in other blood cancers has benefited from the use of patient-derived xenografts, where human leukemias are induced by injecting cells from patients into immunocompromised mice. Historically, MPN xenotransplantation has proven problematic in regular immunodeficient strains. Speculation that this was due to the absence of species-specific growth factors led to efforts to “humanize” the host environment. Along these lines, the Majeti group demonstrated that “ossicles” generated from human bone marrow stroma and ectopically implanted in the flank of mice can be robustly engrafted by malignant cells, including myelofibrosis cells.¹  Although technically challenging, this system includes a human bone marrow–like environment and is readily accessible (e.g., for serial sampling). Other groups have explored mouse strains such as MISTRG mice expressing human mSCF, IL-3, GM-SCF, TPO, and signal regulatory protein alpha (SIRPα) on a Rag2/Il2rγ-deficient genetic background.²  While engraftment of myelofibrosis in these mice was substantially improved, induction of fibrosis was inconsistent.² 

In a recent article in Cancer Discovery, Dr. Hamza Celik and colleagues elegantly and methodically optimized an approach for engrafting myelofibrosis CD34⁺ stem/progenitor cells from peripheral blood of patients into immunodeficient mice called NSG-S that express human IL3, SCF, and GM-CSF. Virtually all mice transplanted with more than 100,000 cells engrafted. Intraosseous delivery of cells was significantly more successful than intravenous inoculation. This suggests a defect in the ability of the injected cells to migrate to marrow, although once engrafted, myelofibrosis cells successfully engrafted the contralateral limb and the spleen.

Most samples caused a lethal disease including marked reticulin fibrosis, splenomegaly, and anemia without acute leukemia, which was remarkably similar to the spectrum of human disease. This enabled genetic and pharmacologic validation of putative targets, and novel drug combinations of JQ1, a BET bromodomain inhibitor, or a PIM kinase inhibitor together with ruxolitinib, a widely used JAK1/2 inhibitor, reduced myeloproliferation and ameliorated bone marrow fibrosis in the mice.

The mutational landscape of the patient samples was maintained in xeno-transplanted mice. The team was able to tease apart clonal architecture such as the evolution of JAK2 plus TP53-mutant MPN on a background of TET2/DNMT3A clonal hematopoiesis. They also found that rare leukemic subclones driven by mutations in both EZH2 and TP53, although clinically “silent” at the time of sampling from the patient, manifested in the xenografts. Strikingly, the outgrowth of leukemogenic clones in the xenografts predicted a progression to sAML that occurred several years later in the patient.