Clonal Heterogeneity in Differentiation Response and Resistance to the IDH2 Inhibitor Enasidenib in Acute Myeloid Leukemia

Background

Mutations in Isocitrate Dehydrogenase 2 (IDH2) occur in many cancers including Acute Myeloid Leukemia (AML). In preclinical models mutant IDH2 (mIDH2) causes partial hemopoietic differentiation block¹. Recently, we showed that single agent enasidenib, a first-in-class, selective mIDH2 inhibitor, produces a 40% response in relapsed/refractory AML patients by promoting differentiation². Here, we studied response and acquired resistance to enasidenib, in sequential samples treated in the Phase 1 study of Enasidenib in relapsed/ refractory AML patients.

Results

We studied a cytogenetically and genetically representative subset of 25 patients enriched for enasidenib responders, genotyped by whole exome sequencing (WES) or cancer gene panel targeted re-sequencing. Pre-enasidenib, differentiation arrest in these AML patients resulted in abnormally expanded leukaemic progenitors or precursors and diminished mature haematopoietic populations. Complete remission (CR) post-enasidenib was associated with in increased mature populations, near-normalisation of haematopoietic progenitor profiles, and restoration of in vitro progenitor function. In most patients, mature blood cells (of erythroid and granulocyte-monocyte lineages) post-enasidenib are IDH2 mutant, consistent with enasidenib inducing differentiation of IDH2 mutant leukaemic progenitors/ precursors.

Each mIDH2 patient studied had on average 13 somatic, non-synonymous exonic or splice site mutations in addition to IDH2 . We used single cell genotyping (SCG) to reveal linear or branching clonal structures in mIDH2 AML. We combined clonal structure data and immunophenotyping of haematopoietic progenitor, precursor and mature populations to track functional behaviour of mIDH2 clones before, and during enasidenib treatment. We demonstrate, for the first time, that mIDH2 subclones within the same patient are functionally heterogeneous: both in their ability to differentiate pre-enasidenib, and in their sensitivity to Enasidenib-induced differentiation. This suggests that different combinations of co-operating mutations result in functional heterogeneity of mIDH2 clones. When we studied the contribution of mIDH2 clones to functional haematopoiesis at CR, we found that this was supported by either ancestral or leukaemic terminal mIDH2 clones.

Despite a median survival of 18-21 months in patients who respond to enasidenib, most patients eventually relapse³. In contrast to targeted therapies such as tyrosine kinase inhibitors, in all 12 relapse samples studied, none had second site mutations in IDH2 . Furthermore, 2-hydroxyglurate (2HG) levels remain suppressed in most patients suggesting enasidenib remains effective in inhibiting mIDH2 enzyme. Instead, mIDH2 clones, which had persisted at CR or partial remission (PR) acquired additional mutations or aneuploidy, highlighting bypass pathways which re-impose differentiation arrest. We found 4 patterns: i) acquisition of IDH1 codon R132 mutations which resulted in a rise in 2HG (n=2), ii) deletion of chromosome 7q (n=4), iii) gain of function mutations in genes implicated in cell proliferation (FLT3, CSF3R) (n=3) and iv) mutation in hematopoietic transcription factors (GATA2, RUNX1) (n=2). We also found mutations in 4 genes (DHX15 and DEAF1 (n=1) ; NFKB1 (n=1) and MTUS1 (n=1)) not previously implicated in haematopoietic differentiation arrest which were selected for, or evolved in mIDH2 subclones at relapse.

Conclusion

This study provides a paradigm of how deep clonal single cell analysis in purified hemopoietic compartments in sequential samples through therapy reveals clonal complexity and the impact of the selective pressure of therapy on clonal architecture. Furthermore, we gain insights into the functional heterogeneity of mIDH2 subclones in their ability to differentiate pre-and post-Enasidenib. Further analysis of this kind in a larger cohort of IDH2 -inhibitor-treated patients would also provide insight to improve efficacy of this novel class of therapeutics, and design of combination therapies in AML and other cancers. Finally, this provides a platform for further study of the pathways mediating enasidenib resistance.

References

1. Kats, L.M. , et al. Cell Stem Cell14, 329-341 (2014).

2. Amatangelo, M.D. , et al. Blood (2017).

3. Stein, E.M. , et al. Blood (2017).