Discovering the drivers of clonal hematopoiesis

Oriol Pich, Iker Reyes-Salazar, Abel Gonzalez-Perez, Nuria Lopez-Bigas

Preprint posted on 23 October 2020

Article now published in Nature Communications at

Repurposing tumour-blood samples to study Clonal Hematopoiesis

Selected by Irepan Salvador-Martinez


Blood cells are generated throughout the life of every person. The process of producing blood cells (i.e. hematopoiesis) starts in the bone marrow, where stem cells divide to give rise to hematopoietic stem cells. Every person has a limited amount of hematopoietic stem cells (HSCs) that, as any other cell in our body, accumulate random mutations with time. Most of these mutations are expected to be neutral, but in some cases they confer an advantage to the HSC progenitors, so that their clonal progeny is over-represented in the total population of blood cells.

This phenomenon is called clonal hematopoiesis (CH) and can be identified by sequencing the DNA of blood cells: if a relatively large proportion of blood cells have a specific cell mutation, these are assumed to be a clonal population derived from a single HSC progenitor. Although this seems to be a normal phenomenon related with age, it has been also associated with risk of hematological cancer and death [1].

Recurrent mutations in specific genes have been found in CH suggesting these might be driver mutations, i.e. they provide a selective advantage to the cells leading to clonal expansion. A complete characterisation of gene specific mutations that might drive CH would be useful to investigate the mechanisms that lead to CH in the first place. Pich et al. tackle this problem by identifying signals of positive selection in blood somatic mutations, similar to cancer research studies.

About the preprint

For identifying positive selection in blood somatic mutations, the authors employed a clever approach, repurposing paired blood-tumour samples of more than 12,000 donors, divided in primary (N=~ 8,000; whole-exome level) and metastatic (N=~4,000; whole-genome level) cohorts.
These datasets were originally used to analyse mutations in the DNA sequence of the tumour sample, using the blood sample as control (variant calling). In this study, they analysed the mutations in the blood cells, using the tumour sample as control (reverse variant calling).

On the many genome somatic mutations obtained by reverse calling in both cohorts, they applied the IntOGen bioinformatic pipeline, which analyses the mutation data with seven different driver discovery algorithms combining the results to identify signals of positive selection [2]. Signals of positive selection can be: unexpected high recurrence of mutations; unexpected clustering of mutations in certain regions of the gene; or exceptionally high functional impact of the observed mutations.

After the IntOGen pipeline the authors filtered out few genes deemed as possible artifacts to reduce the probability of having false positives and kept only genes with previous evidence of being associated with CH, myeloid malignancies, or tumorigenesis in general, leading to 26 and 23 driver candidate genes in the metastasis and the primary cohorts, respectively (Fig 1).

Figure 1 (a) Summary of the discovery analysis applied to blood somatic mutations detected across the primary and metastasis cohorts. The somatic mutations identified across all donors of a cohort were analysed with the IntOGen pipeline to identify different signals of positive selection. (b) CH driver genes discovered across the primary and metastasis cohorts. Genes known to be involved in CH, myeloid malignancies or tumorigenesis in general are labeled. (from Figure 2 in the preprint made available under a CC-BY-NC-ND 4.0 license).


Importantly, most genes already associated with CH were recovered, as well as previously found associations such as, CH being positively influenced by age and by the exposure to cytotoxic treatments. Interestingly, although a set of CH genes common to both cohorts was found, many genes were specific to each cohort. The authors point out that this might be due to differences in treatments (e.g. chemoterapy) ethnicity or lifestyle exposures between the donors of both cohorts. Mutations in some CH-related genes are indeed known to provide an advantage to hematopoietic cells under exposure to certain cytotoxic treatments.

The authors then enriched these datasets with a “targeted-cohort”, consisting of N=~24,000 paired blood-tumour samples where a subset of all protein coding genes (N=468) have been sequenced (mostly genes involved in tumour development). Interestingly, they showed that only 44 genes show positive selection signatures. Other genes appeared mutated but didn’t show selection signatures, suggesting these are just passenger mutations and do not drive CH.

This study shows the utility of repurposing massive datasets of paired blood-tissue samples for the study of CH and represents a step towards a targeted sequencing panel of CH drivers that could be widely used for screening and discovery of early stages of CH in patients.

Questions to the authors

  • In some types of cancers, studies have shown some mutations seem to occur as a recurrent series of events. Do you think this could also be the case for CH driver genes?
  • Is there any evidence for differences in CH drivers for different human populations?


[1] Genovese G, Kähler AK, Handsaker RE, et al. Clonal Hematopoiesis and Blood-Cancer Risk Inferred from Blood DNA Sequence. N Engl J Med. 2014;371(26):2477-2487. doi:10.1056/nejmoa1409405
[2] Martínez-Jiménez F, Muiños F, Sentís I, et al. A compendium of mutational cancer driver genes. Nat Rev Cancer. 2020;20(10):555-572. doi:10.1038/s41568-020-0290-x


Posted on: 23 November 2020


Read preprint (No Ratings Yet)

Author's response

Nuria Lopez-Bigas and Abel Gonzalez-Perez shared

Thanks very much for highlighting our manuscript and for the clear summary of our work.

• In some types of cancers, studies have shown some mutations seem to occur as a recurrent series of events. Do you think this could also be the case for CH driver genes?

Indeed in some cases cancer evolution has been shown to occur through an ordered series of events. This was observed, for instance, many years ago in colorectal cancer by Bert Vogelstein and colleagues (mutations in APC, KRAS, TP53). And it has also been shown that in the case of some hematological malignancies the order in which mutations occur has a profound impact on disease evolution.

In the case of clonal hematopoiesis we have seen that in most of the 12.000 cases we have analyzed only one mutation affecting a CH gene is identified. But in 18% of the cases we identify more than one and we can observe some preferences of co-occurring mutations (Fig. 4e). But more data is needed to discern whether a series of mutations cooperate in the development of CH and this is modulated by the order in which they occur.

• Is there any evidence for differences in CH drivers for different human populations?

That is also a very interesting question. There are differences in the CH drivers depending on exposures to different agents and conditions. For example, as we explore in the paper, across patients who have received certain chemotherapies there is an enrichment of PPM1D mutations. Other associations between mutations affecting specific CH drivers and lifestyle exposures have also been reported (e.g. smoking with preference ASXL1). There are also germline variants associated with different risks to develop CH. In that respect it is likely that there are differences in CH drivers for different human populations, due to differences in genetic background. In addition, differences in exposures related to varying lifestyles across populations may also modulate the selective advantage provided by particular mutations affecting HSCs.

Have your say

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here

Also in the cancer biology category: