A new gene expression atlas developed using single-cell RNA sequencing data shows how normal hematopoietic cells differentiate and was used to catalogue the multiple ways that aberrant differentiation can lead to acute myeloid leukaemia (AML), according to results presented at the American Association for Cancer Research (AACR) Annual Meeting 2025, held April 25-30.

This study and an accompanying commentary are being simultaneously published in the AACR journal Blood Cancer Discovery.

The comprehensive dataset may shed light on how AML driver genes affect cell differentiation in different contexts, which may inform new biomarkers and drug targets.

It also provides a toolkit with which AML researchers can interrogate their own data to address new hypotheses.

“This represents an important step in evaluating cell state heterogeneity within and across AML patients in a more quantitative way, adding a new tool to the toolkit for AML researchers,” said Andy Zeng, PhD, an MD/PhD student at the University of Toronto, who presented the study.

AML is characterised by extensive heterogeneity that can hinder the effective prediction of prognosis and response to therapy, Zeng explained.

He emphasised that not all of this heterogeneity is driven by genetic mutations, and that different types of leukaemia cells can exist within an individual patient’s disease.

He believes that a more thorough understanding of gene expression changes during the development and progression of AML can help researchers better define and target different types of AML cells.

However, to understand how haematopoiesis goes wrong to cause AML, researchers must first characterise normal haematopoiesis.

However, building a precise and comprehensive reference map of normal haematopoiesis has been a significant technical challenge for the field, according to Zeng.

“The purpose of this study was to establish an authoritative reference of haematopoiesis and to use that reference to comprehensively catalogue all the different types of leukaemia cells present in AML,” Zeng said.

Zeng and colleagues—including the study’s senior author John E. Dick, PhD, FAACR, a senior scientist at the Princess Margaret Cancer Centre and a member of the AACR Haematologic Malignancies Working Group Steering Committee—assembled a reference atlas of normal human haematopoiesis using gene expression data from 263,519 individual cells enriched for haematopoietic stem and progenitor cells (HSPCs).

Upon establishing this atlas, the researchers mapped the transcriptional profiles of more than 1.2 million cells from 318 patients with leukaemia to distinct cell states.

The cohort chiefly consisted of patients with AML but also included patients with mixed-phenotype acute leukaemia (MPAL) and acute erythroid leukaemia (AEL), two rare leukaemia types that have some overlapping characteristics with AML.

The researchers identified at least 12 distinct differentiation patterns across the AML samples.

Some of these patterns closely resembled MPAL or AEL, highlighting how patients with distinct disease diagnoses can nonetheless share very similar differentiation anomalies.

These similarities emphasise the ambiguity in the diagnostic boundaries between the disease classes, Zeng said.

Zeng and colleagues further sought to characterise the genetic underpinnings of these differentiation patterns.

Using bulk RNA sequencing data from more than 1,200 AML patient samples, they estimated the abundance of each leukaemia cell state in each patient sample and correlated these cell states with common genetic alterations.

They found that the same mutation could shift cells in different directions depending on the cell of origin and/or on other co-occurring mutations.

In this way, a single leukaemia-initiating genetic alteration can generate leukaemias with different biological properties, Zeng explained.

“These findings reveal that the phenotypic heterogeneity in AML arises from the interplay between genetic drivers and the specific cellular context, helping us begin to decipher the ‘rules’ governing the factors that shape the disease,” Zeng said.

Ultimately, the data offer a toolkit with which researchers can interrogate their own AML samples to find detailed associations between cell states and genetic or clinical variables with potential therapeutic applications.

Zeng hopes that utilising these tools will uncover clinical biomarkers of these cell states that could better predict prognosis and facilitate precision medicine for patients with AML.

“By establishing a high-resolution single-cell reference atlas of haematopoiesis, we’ve not only advanced our understanding of how differentiation goes awry in AML, but we’ve also provided an accessible toolkit for other researchers to rapidly map and classify their own blood samples profiled by single-cell RNA sequencing,” Zeng said.

“We also hope this motivates others outside of the AML space to adopt similar analytical frameworks that precisely compare cancers with their normal tissue counterparts, as these approaches may allow them to integrate genetic and cellular models of cancer heterogeneity in order to advance precision medicine across oncology.”

Limitations of this study include technical issues associated with single-cell RNA sequencing, limited sensitivity to detect extremely rare cell types, and a markedly heterogeneous cell population that could complicate the observed associations between genetic drivers and cell states.

Funding for this study was provided by the University of Toronto, the Princess Margaret Cancer Foundation, the Ontario Institute for Cancer Research, the Canadian Institutes of Health Research, the International Development Research Centre of Ottawa, Canada, the Canadian Cancer Society, the Terry Fox New Frontiers Programme, the University of Toronto Medicine by Design Initiative with funding from the Canada First Research Excellence Fund, the Ontario Ministry of Health, the Canada Research Chairs Programme, the American Lebanese Syrian Associated Charities of St. Jude Children’s Research Hospital, the Alex’s Lemonade Stand Foundation for Childhood Cancer, the National Cancer Institute of the National Institutes of Health, St. Baldrick’s Foundation, and the Henry Schueler 41 & 9 Foundation.

Zeng declares no conflicts of interest.

Source: American Association for Cancer Research