Subclonal “skewing” Of De Novo AML Samples After Engraftment In Immunodeficient Mice

Most recurrent somatic mutations in acute myeloid leukemia (AML) have been identified. However, it is now clear that many mutations are not always present in all the cells within an AML sample. Our previous studies have shown that all AML samples are comprised of a founding clone and usually one or more subclones that are derived from the founding clone. The clonal architecture of an AML sample can be identified using single nucleotide variants (SNVs) that cluster according to discrete variant allele fractions (VAFs); the accurate identification of these clusters in AML samples generally requires deep digital sequencing of all the variants identified by whole genome sequencing (WGS), since AML samples have so few mutations in coding sequences. However, it is not yet clear whether human AML samples engrafted in immunodeficient mice accurately recapitulate the subclonal architecture of the injected sample. Similarly, it not yet clear whether the presence of human cytokines, such as those expressed in the NSG-SGM3 strain (NSG mice with transgenes expressing human IL-3, SCF, and GM-CSF) would alter the engrafting potential of individual subclones within an injected sample.

We injected 1 million bulk (unsorted and unmanipulated) cells from 9 different oligoclonal, de novo AML samples into 73 mice (31 NSG and 42 NSG-SGM3) via lateral tail vein injections. Five of the 9 samples were concurrently injected into both NSG and NSG-SGM3 mice. Engraftment was then assessed at 14 weeks (or the first sign of illness) using flow cytometry for human CD45, CD33 and/or CD34. Eight of the 9 samples engrafted, and 58 of the 73 mice had detectable leukemia at 14 weeks (4-11 mice per engrafted AML sample). Consistent with previous studies, engraftment was more efficient and robust in the NSG-SGM3 strain: 39/42 (92.9%) NSG-SGM3 mice had greater than 1% human AML in the bone marrow, compared to 19/31 NSG (61.3%). NSG-SGM3 mice had an average 47.5% human AML cells in the bone marrow, compared to 23.9% in NSG mice (P<0.05).

We performed targeted deep sequencing on human AML cells purified from the engrafted mice to define the VAFs of all known somatic variants previously identified by whole genome sequencing for these cases. We have evaluated 36 xenografts from 5 different AML samples; 3 of these samples were injected into both NSG strains. Despite multiple subclones present per sample at the time of injection, 31 of 36 xenografts only contained cells from a single subclone. 3/5 samples consistently engrafted the same subclone, while multiple different subclones engrafted from the other two samples. Preferential engraftment and outgrowth of subclones that comprised <20% of the input sample was observed for 2 of the samples, implying that these subclones have cell-intrinsic properties that provide them with a growth or engraftment advantage. Mutations in DNMT3A (present in 3 AMLs and 19 xenografts) and NPM1 (present in 2 AMLs and 15 xenografts) were always retained in the xenografts. In contrast, variable retention of FLT3-ITD was observed--for 2 samples the subclone bearing the FLT3-ITD engrafted in all xenografts (n=4 for each AML) while the xenografts from another AML sample with a subclonal FLT3-ITD allele never contained this mutation. Lastly, comparison of the xenografts from the same AML sample in NSG and NSG-SGM3 mice demonstrated variable patterns: the xenografts from some AML samples were comprised of the same subclone in both strains, but from another sample, the two strains produced xenografts with different clonal compositions. Thus, the presence of human cytokines can skew subclonal engraftment and/or proliferation in immunodeficient mice.

In summary, engraftment of AML samples in NSG mice (and NSG-SGM3 mice) typically results in skewing of the sample’s original clonal architecture with a dramatic restriction in the number of subclones. Further, the expression of human hematopoietic cytokines can influence the engraftment and/or outgrowth of specific subclones. These findings suggest that rigorous analyses of samples before and after xenotransplantation are necessary to define the subclones (and the mutations contained therein) for the testing of novel targeted therapies.