A Murine Model of Human Primary Leukemia for Studying Leukemia-Stromal Interactions and Prediction of Clinical Outcome.

Acute or chronic leukemias resist apoptosis in vitro when co-cultured with marrow stromal cells, suggesting that the growth/survival of leukemia cells relies in part on interactions with stromal cells in the microenvironment. We have recently demonstrated that consistent and high-level engraftment of human primary leukemia obtained from patients can be achieved in NOD/scid mice by preconditioning with either adherent cord blood or marrow mesenchymal stem cells. High success rate of engraftment (84.9 ± 2.9 % leukemia blasts in mouse marrow) was obtained with many lineages of leukemia including acute T- or B-cell lymphoblastic and myeloid leukemia. In general, leukemia blasts were detectable in peripheral blood of mice by week 4–5, leading to fatal outcome by week 6 (42 ± 4 days). Furthermore, cells from the marrow of these preconditioned mice were found to secrete leukemia-promoting activities, suggesting that the mouse marrow had been altered to favor the proliferation/survival of leukemic cells in vivo. We also showed that the human leukemia cells harvested from mice could be serially transferred to other mice for many generations with ~100 fold increase in the number of leukemic and clonogenic cells in mice, while retaining properties similar to primary leukemia samples obtained from patients. Several lines of evidence in our studies including the appearance of leukemia blasts in marrow and blood, the dissemination to other tissues, and gene expression profiling in microarray analysis confirm that this xenograft model recapitulates key features of human leukemia. Furthermore, weekly i.p. injections of NOD/scid mice with vincristine at 0.5 mg/kg for three weeks, starting at second week after inoculation of patients primary leukemia, resulted in a significant delay in the appearance of human leukemia blasts in blood, doubly the length of mouse survival. In addition, pairs of primary leukemia samples collected at diagnosis and at relapse from the same patients were engrafted into NOD/scid mice. The NOD/scid mice transplanted with either samples developed leukemia in mouse peripheral blood at week 4–5 with similar kinetics after inoculation. However, weekly vincristine treatment x 3 of the mice transplanted with diagnosis leukemia samples, prevented the appearance of leukemia blast cells in the circulating peripheral blood for at least 5 weeks. In contrast, similar treatment of mice engrafted with relapse leukemia samples had 50% leukemia blasts in blood at week 5 and subsequently developed fatal leukemia dissemination at week 7. These findings indicate that this robust high level engraftment model of human primary leukemia may be useful in predicting clinical response to chemotherapy. Evidence has further suggested that these human leukemia in NOD/scid mice may be derived from a small subset of immature stem cells in the samples that give rise to leukemia expansion and phenotypic diversity in these mice. Therefore, we had developed a robust and predictive animal model of human primary leukemia, which is valuable for studying leukemia stem cells and for testing or prioritizing new agents/regimens in vivo.