A transcriptomic continuum of differentiation arrest identifies myeloid interface acute leukemias with poor prognosis

“Is it time to consider AML treatment for poor-risk immature T-ALLs? We may have transcriptomic data that indicates just this”.
A transcriptomic continuum of differentiation arrest identifies myeloid interface acute leukemias with poor prognosis

I had just started my laboratory in Cambridge when Elizabeth Macintyre and Jonathan Bond from the Necker hospital in Paris came knocking at my email inbox with this question. I am a stem cell biologist and my research aims to understand how all blood cell types are made from haematopoietic stem and progenitor cells in humans, with a focus on transcriptomics. I am a firm believer that there is lots to be learned from comparing tumours to the normal cellular hierarchies that they hijack, and so I embarked on this collaboration.

To date, leukaemias are split into either the lymphoid or myeloid type at diagnosis based on leukemic cell morphology and expression of lineage-specific surface markers. Correct diagnosis is key to give patients the best chance of survival. But in some “immature” cases, the leukaemic cells express markers of several lineages or none, posing a challenge for treatment allocation. Elizabeth, Jonathan and their colleagues from French acute leukemia cooperative groups had put together a cohort of samples from T- acute lymphoblastic leukaemias (T-ALLs) and acute myeloid leukemia (AML) patients, intentionally enriched for these “immature” types. They noticed that at the transcriptional level, some T-ALLs clustered together with AMLs. Together we decided to investigate the significance of this original finding and its potential relevance to therapy.

We employed a number of bioinformatic methods, but the most interesting insights came from Iterative Clustering and Guide gene Selection (ICGS), a tool initially developed to cluster single cells into states recapitulating differentiation trajectories 1. Within our dataset, it identified a series of T-ALLs and AMLs that did not express the classical features of either T-ALL or AMLs, specifically lacking T cell or myeloid cell identity. We found that these “interface” T-ALLs and AMLs were not restricted to previously described categories of immature leukemias (such as ETP-ALL or M0-AMLs) or specific mutations, indicating that interface attributes transcend current phenotypic/genotypic classifications and represent a new entity.

Interface Acute Myeloid Leukemias include cases with early B lymphoid differentiation orientiation and poor response to AML therapy.

The most exciting data came from interface AMLs. When we developed a transcriptional interface AML signature (IAL), this unexpectedly strongly correlated with poor prognosis in 2 independent AML patient cohorts, independently of other common prognostic factors. Interestingly, a leukemic stem cell (LSC) signature developed by the groups of Drs Jean Wang and John Dick in Toronto is making its way into clinical trials 2,3, underscoring the clinical potential of transcriptional signatures for leukemia treatment allocation. Importantly, our IAL signature did not overlap with the LSC signature. The latter reports stemness, intended as capacity to initiate and maintain the tumour. We speculate that, akin to healthy HSCs, distinct transcriptional modules may control self-renewal and absence of lineage affiliation in acute leukemia. We also found that AML patients with high IAL scores responded poorly to Gemtuzumab Ozogamycin, currently used in clinical protocols in combination with Daunorubicin and AraC. Unexpectedly, in these poor responders the transcriptomes expressed high levels of genes associated with B cell differentiation. We therefore suggest that AMLs with high IAL scores may benefit from ALL-type regimens, using glucocorticoids, as recently described for RUNX1-mutated AMLs 4, all identified as IAL in our study.

So, is it time to consider AML treatment for poor-risk immature T-ALLs? We still do not know and I am looking forward to exploring this in the next chapter of this rewarding trans-european collaboration. But to the question: “Is it time to consider ALL treatment for some poor-risk AMLs?”, we would now definitely answer yes.



  1. Olsson, A. et al. Single-cell analysis of mixed-lineage states leading to a binary cell fate choice. Nature 537, 698–702 (2016).
  2. Ng, S. W. K. et al. A 17-gene stemness score for rapid determination of risk in acute leukaemia. Nature 540, 433-437 (2016).
  3. Bill, M. et al. Mutations associated with a 17-gene leukemia stem cell score and the score’s prognostic relevance in the context of the European LeukemiaNet classification of acute myeloid leukemia. Haematologica 105, 721–729 (2020).
  4. Simon, L. et al. Chemogenomic Landscape of RUNX1-mutated AML Reveals Importance of RUNX1 Allele Dosage in Genetics and Glucocorticoid Sensitivity. Clin. Cancer Res. Off. J. Am. Assoc. Cancer Res. 23, 6969–6981 (2017).

Please sign in or register for FREE

If you are a registered user on Nature Portfolio Cancer Community, please sign in