The aim of this study was to determine the influence of the engraftment site on the tumor biology and drug sensitivity of a panel of hematological patient derived xenografts (PDX). PDX cells (3x10e6 cells/mouse) were injected intratibially (i.t.), intrasplenically (i.s.) or subcutaneously (s.c.) into NOG (NOD/Shi-scid/IL-2Rγnull) mice. Leukemic cell engraftment was determined by flow cytometry (FC) in bone marrow (BM), peripheral blood (PB) and spleen during the course and at the end of the experiment. Overall survival (OS) served as an additional read-out. In nine models, sensitivity towards cytarabine (Cy) was evaluated across the different implantation sites. Up to now our group has established 26 PDX of acute leukemia (23 AML, one ALL, two APL) covering a broad range of the different genetic subtypes: amongst other molecular alterations models carrying PML-RARA or AML1-ETO fusion proteins are included in the collection. 24/26 lines engrafted when injected i.t., 17/22 developed tumors after i.s. implantation and 16/19 established tumors post s.c. injection of T cell depleted PB or BM of patients. Hence, for the majority of the models the engraftment capacity per se was not dependent on the injection site. Yet, some models could exclusively be propagated in a specific setting: one AML model, grew solely i.t. or i.s., whereas another AML and one APL could be propagated only when injected s.c.. The implantation site did influence tumor growth behavior: Mean OS ranged from 161.3 (±22.3) days for i.t. to 93.1 (±13.1) days for s.c. propagation. I.s. transplanted mice had to be sacrificed after 137.6 (±25.2) days. The differences in mean OS between the three settings were not statically significant (one way ANOVA). Nevertheless, the OS times within single models differed significantly depending on the implantation technique (p ranging from < 0.005 - 0.017, Log-rank (Mantel-Cox) test). The dissemination pattern of individual lines was as well affected by the injection site. In general, infiltration of the investigated hematopoietic organs was significantly higher when cells were engrafted i.t. or i.s as compared to the s.c. approach (p< 0.0001, one way ANOVA). The mean BM (i.t.: 58.1 ± 8.1%; i.s.: 51.3 ± 13.4%; s.c.: < 0.01%), spleen (i.t.: 40.4 ± 8.8%; i.s.: 52 ± 11.5%; s.c.: < 0.01%) and PB (i.t.: 36.6 ± 8.1%; i.s.: 36.4 ± 12.0 %; s.c.: 10.4 ± 7.2 %) infiltration rate across all 26 models was similar in the i.t. and the i.s. setting but considerably higher than in the s.c approach. Significant differences could be detected between individual models depicting characteristic engraftment patterns independent of the injection site (p ranging from < 0.003 - 0.02, one way ANOVA). The expression pattern of the six investigated surface markers (CD45, CD3, CD34, CD33, CD38 & HLA-ABC) was not influenced by the application route. Every model depicted its distinct expression pattern similar to the expression pattern of the donor sample. Of note, the expression pattern was stable across different passages in the murine host. Cy was highly active in five AML and one ALL model: OS was significantly prolonged in the s.c. as well as in the disseminated setting (p ranging from < 0.001 - 0.003, Log-rank (Mantel-Cox) test). Three other AML models depicted a less pronounced sensitivity towards Cy (p ranging from < 0.005 - 0.007, Log-rank (Mantel-Cox) test) both growing s.c. or i.t.. Thus, drug sensitivity was not influenced by the injection site of the leukemic cells. To the best of our knowledge, this is the first time that such a thorough side-by-side comparison was performed in acute leukemia PDX models. Taken together, the application route has a major impact on the tumor biology of the PDX model: i.t. and i.s but not s.c. implantation induced a disseminated growth pattern mimicking consistently the human disease. Up to now, implanting leukemic cells i.t. led to the highest engraftment rate (92%) within the panel of established AML PDX models, indicating the importance of the hematopoietic niche as a supportive tumor microenvironment. Therefore, the establishment of favorable genetic AML subtypes should be preferably done by i.t. implantation. Nevertheless, the consistent expression of surface markers as well as sensitivity towards Cy independent of the implantation site, justifies the use of s.c. implanted AML PDX for screening approaches in an early stage of the drug development process.


Schueler: Charles River Research Services Germany GmbH: Employment. Lenhard: Charles River Research Services Germany GmbH: Employment. Klingner: Charles River Research Services Germany GmbH: Employment. Oswald: Charles River Research Services Germany GmbH: Employment.

Author notes


Asterisk with author names denotes non-ASH members.