Single-Cell Proteogenomic Analysis of Clonal Evolution in Patient-Derived Xenograft Models of AML Treated with IDH Inhibitors

The clonal architecture of acute myeloid leukemia (AML) can be highly complex and heterogeneous. This genetic diversity complicates the treatment of AML, as different clones may respond variably to therapies, contributing to treatment resistance and disease relapse. Understanding the mechanisms and implications of clonal evolution in AML is crucial for developing more effective treatment strategies that can target the diverse cellular populations.

The generation of patient-derived xenograft (PDX) models has revolutionized the study of AML. One of the main uses of PDX models is to study how a leukemic sample might respond to different therapies. Here, we used single-cell DNA sequencing (scDNA-seq) and surface protein profiling to monitor the clonal evolution and differentiation of an isocitrate dehydrogenase 1 (IDH1)-mutated leukemic sample in response to ivosidenib (IVO), a mutant IDH1 inhibitor, either as monotherapy or in combination with venetoclax (VEN) or azacitidine (AZA). Single-cell mutation analysis of the engrafted cell population showed that the initial evolutionary path was linear characterized by the sequential acquisition of DNMT3A^R882H (clone 1), WT1^R467W (clone 2), IDH1^R132H (clone 3), NPM1^{Trp288Cysfs*12} (clone 4), FLT3-ITD (clone 5), and WT1^R439C (clone 6) mutations. Three subclones branched off at the end of this linear path with each acquiring either a TET2^Q1548del (clone 7), DNMT3A^D531del (clone 8), or KIT^H40QfsTer6 (clone 9) mutation. The overall clonal composition of the input primary sample and sample from vehicle-treated animals were largely similar. To quantify the differences in therapeutic response between clones, we developed a hierarchical multinomial Bayesian model that estimated the fold change in absolute cell numbers for each clone compared with vehicle treatment. This analysis revealed that both single-agent VEN and IVO+VEN were generally more effective in targeting earlier-stage clones (clones 3-5) than later-stage clones (clones 6-9). In contrast, single-agent AZA and IVO+AZA were effective in eliminating clone 6 as well as earlier-stage clones. Notably, the combination of IVO+AZA was effective in reducing all the clones by approximately the same magnitude, including clones 7 and 8 which demonstrated lower sensitivity to AZA monotherapy. Given that the WT1^R439C mutation, acquired in clone 6, distinguished early-stage from late-stage clones, these findings suggest that WT1 mutations might contribute to IVO and VEN resistance and that the addition of AZA can overcome this mechanism of resistance. Consistent with these findings, the combination of IVO+AZA strongly upregulated the expression of the myeloid markers CD11b, CD14, and CD15 to a greater extent than with either single agent. These findings provide evidence that AZA and IVO could synergize to overcome the differentiation block in IDH1-mutated AML cells, reflecting the superior clinical efficacy of this regimen over single-agent IVO.

To broaden the applicability of PDX models to study the competition between genetic clones that co-exist infrequently in an individual sample, we generated mixed PDX (mPDX) models by co-engrafting two or more leukemic samples in the same animal and used scDNA-seq to deconvolute the clonal composition and sample origin of the engrafted cells. As proof-of-principle, we generated an IDH1^R132H/IDH2^R140Q mPDX to model isoform switching in both directions by treating the animals with either IVO or enasidenib (ENA), a mutant IDH2 inhibitor. We observed the expected differentiation response and depletion of IDH1-mutated or IDH2-mutated clones by IVO or ENA treatment as single agents, respectively. Further studies are underway to demonstrate whether dual mutant IDH1 and IDH2 inhibition (IVO+ENA) will circumvent resistance driven by isoform switching.

In summary, we demonstrate the utility of applying single-cell proteogenomic analysis in traditional and mixed PDX models to gain crucial insights into mechanisms of resistance and potential strategies to overcome it.