Prevalent premalignancy

Adam S. Sperling and Benjamin L. Ebert

The presence of hematologic malignancy–associated mutations in the blood of individuals without cytopenias or dysplasia has been termed clonal hematopoiesis of indeterminate potential (CHIP), a clinical entity associated with increased age, risk of developing hematologic malignancy, and decreased overall survival.1 Two papers in this issue of Blood take different approaches to advance our understanding of the frequency and clinical characteristics of clonal hematopoiesis. Zink et al analyze whole genome sequencing data from the blood of >11 000 healthy Icelandic individuals, and Buscarlet et al perform targeted sequencing on the blood of 2530 healthy older women.2,3

The presence of clonal hematopoiesis was first inferred from the finding that X-chromosome inactivation, predicted to be a purely stochastic event, is often skewed in the blood of otherwise healthy women, a subset of whom were later shown to harbor mutations in the TET2 gene.4 In addition, hematopoietic stem cells (HSCs) carrying preleukemic mutations can be detected in both leukemia and remission samples obtained from patients.4 More recently, the presence of somatic variants in the blood of healthy individuals has been used to identify and characterize clonal hematopoiesis. Analysis of exome sequencing data from >30 000 people identified recurrent somatic hematologic malignancy–associated mutations in up to 10% of patients over the age of 65.5-7 Together, these findings have led to a model in which HSCs acquire mutations over time leading to clonal expansion and eventually to overt malignancy.8

Most somatic mutations are biologically silent passenger events that create a mutational signature marking each HSC and its progeny.9 The presence of a detectable collection of somatic mutations within the blood is therefore suggestive of clonal expansion from a single HSC. Zink et al take advantage of a passenger mutation analysis to develop an unbiased statistical method to measure outliers based on the number of detectable somatic mutations in order to identify clonal hematopoiesis without depending on the presence of known driver mutations. They find the prevalence of clonal hematopoiesis to be almost double that previously reported within each age group, rising to >50% of those >85 years old. A candidate hematologic malignancy–associated driver mutation could only be identified in ∼40% of these cases. Clonal hematopoiesis, with or without a candidate driver mutation, was associated with increased risk of hematologic malignancy and death. One caveat to their analytic approach is that not all individuals with a detectable somatic driver mutation had enough passenger mutations to meet mutational outlier criteria for clonal hematopoiesis.

Buscarlet et al measure clonal hematopoiesis using a targeted sequencing panel of 19 genes recurrently mutated in myeloid disease. The increased sequencing depth associated with a targeted approach likely explains the increased prevalence of CHIP (16%) seen in their cohort. The use of ultradeep, error-corrected sequencing has previously demonstrated mutations within DNMT3A or TET2 in up to 95% of older adults, suggesting that if you sequence deeply enough you can find a mutant clone in virtually all subjects.10 The most common recurrently mutated genes identified by Buscarlet et al are also DNMT3A and TET2, and these 2 genes account for a somewhat larger proportion (93%) of all mutations than in previous reports.6,7 They further assess the relationship between these mutations and a variety of hematological parameters, demonstrating no significant correlation between mutation status and any abnormality in the blood. This cohort did not have long-term outcome data available for survival analysis.

The high frequency of clonal hematopoiesis observed by Zink et al in the absence of known driver mutations could be explained by the presence of additional undiscovered driver genes or, alternately, by epigenetic alterations leading to clonal expansion in the absence of a somatic driver mutation. The authors do report 1 new candidate driver gene, metastasis-associated protein 2 (MTA2), which encodes a subunit of the NuRD (nucleosome remodeling and histone deacetylase) complex, an epigenetic regulator. Mutations in MTA2 have previously been found in preleukemic HSCs, and many other epigenetic regulators are recurrently mutated in CHIP.8,11 Further work will be needed to confirm whether mutations in MTA2 can, in fact, lead to clonal hematopoiesis and whether there are additional genes that lead to clonal expansion and hematologic malignancy when mutated.

These 2 reports provide evidence of familial predisposition to clonal hematopoiesis. Zink et al note an association between short telomeres and clonal hematopoiesis. They go on to identify a small germ line deletion within the telomerase reverse transcriptase (TERT) intron 3 that is associated with an increased risk of development of clonal hematopoiesis. How this deletion might alter TERT function or lead to increased clonal hematopoiesis is unclear. Because the patient cohort analyzed by Buscarlet et al includes a large number of sibling pairs, they were able to show familial aggregation for TET2-mutation acquisition. A similar familial association has been described in patients with myeloproliferative neoplasms.12

Another question addressed by Zink et al is whether clonal hematopoiesis is always a consequence of a neoplastic driver mutation or whether clonality might, at times, reflect progressive oligoclonal skewing of HSCs during normal aging. In very old individuals, the number of HSCs that actively contribute to mature blood production may be limited, leading to oligoclonality.13 Zink et al develop a computational model to demonstrate that some degree of clonality may, in fact, be an inevitable consequence of neutral drift within the HSC compartment over time. Although clonal hematopoiesis in the presence of a known driver mutation (CHIP) has increased risk of malignant transformation and death, it is unclear whether clonal hematopoiesis due to neutral drift has the same pathological consequences.

Multiple reports have now established that CHIP is a distinct clinical entity defined by a common set of mutations with significant risk for progression to hematologic malignancy and death. Collectively, these 2 papers demonstrate that clonal hematopoiesis is more common than previously reported, may have a genetic predisposition, and is often not associated with an identifiable somatic driver mutation. Future studies are now needed to determine how best to identify patients with clonal hematopoiesis, whether all patients hold the same risk of transformation and death, and how to manage those patients at highest risk.


  • Conflict-of-interest disclosure: The authors declare no competing financial interests.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.