CHIPping away at Erdheim-Chester disease

In this issue of Blood, Cohen Aubart et al report a high prevalence (42.5%) of mutations associated with clonal hematopoiesis (CH) in bone marrow specimens of patients with Erdheim-Chester disease (ECD).¹  These novel findings shed light upon the molecular underpinnings of ECD and its well-known association with other myeloid neoplasms.

ECD was first described in 1930 by William Chester during his visit to Jacob Erdheim in Vienna. However, it was not until 2016 that the World Health Organization finally recognized ECD as a hematopoietic neoplasm. This determination was fueled by the landmark discoveries of MAPK-ERK pathway mutations in most ECD specimens in the last decade, a feat achieved through collaborations facilitated by the ECD Global Alliance. As more data trickled in, it was discovered that ECD patients had a high prevalence of concomitant myeloid neoplasms (∼10%).²  Although several shared mutations have been reported in both entities,^2-4  the pathogenic mechanisms underlying development of myeloid neoplasms in ECD remain elusive. The findings by Cohen Aubart et al provide additional insights into the ontogeny of ECD and its association with myeloid malignancies.

By conducting next-generation sequencing (NGS) for 36 genes on bone marrow specimens of 120 nonconsecutive ECD patients, Cohen Aubart et al identified mutations associated with CH in 42.5%. Most common mutations identified were TET2 (22%) followed by ASXL1 (9%) and DNMT3A (8%). Overall, 18 (15%) patients developed subsequent myeloid neoplasms, the most common ones being myelodysplastic syndrome, myeloproliferative neoplasm, and chronic myelomonocytic leukemia (CMML). Of patients harboring CH, 31% developed a myeloid malignancy vs 3% without CH. Alternately, 89% of patients who developed a myeloid neoplasm had CH vs 34% of those who did not. The investigators flow-sorted a subset of marrow specimens and found similar CH mutations in CD14⁺ monocytes and bone marrow progenitor cells by deep sequencing.

The term clonal hematopoiesis of indeterminate potential (CHIP) is typically distinguished from CH due to its inclusion of primarily myeloid neoplasm-driver genes. In the original proposed definition of CHIP, 19 candidate genes were included, although 3 genes (DNMT3A, TET2, and ASXL1) constitute the major bulk of reported mutations.^5,6  The prevalence of CHIP increases with age, but varies from 10% to 40% among individuals >50 years old, depending on the sequencing technique used, the variant allele fraction (VAF) cutoff, and the health status of population being studied.^5,7  Although the prevalence of CH in the report by Cohen Aubart et al seems quite high, the lack of age-matched control group precludes a quantitative interpretation of the data. Except for the frequencies of TET2, the frequencies of other CH mutations were comparable to those reported in the literature if analyzed by deep sequencing. Therefore, the high frequency of TET2 mutations is the novel finding of this report and was curiously associated with BRAF-V600E mutation and vascular involvement by ECD.

TET2 proteins play a key role in epigenetic regulation of hematopoietic stem cells (HSC) through DNA demethylation. TET2 loss-of-function mutations are associated with DNA hypermethylation and increased risk of hematologic neoplasms.⁵  However, isolated CHIP mutations, such as TET2, DNMT3A, and ASXL1, may not themselves lead to a myeloid malignancy, unless accompanied by an additional genomic “hit,” such as mutations in other myeloid neoplasm driver genes.⁸  The acquisition of other mutations may be random or promoted by the proinflammatory effect of existing CHIP mutations that generates clonal instability.

The findings by Cohen Aubart et al provoke several questions/hypotheses regarding the association of TET2-mutant CH with MAPK-ERK–mutated ECD, and its possible role in the high frequency of concomitant myeloid neoplasms (see figure). Mutations in TET2 (and other CHIP genes) may represent ancestral events, followed by the acquisition of other mutations through chronic inflammation or pressure from the microenvironment (scenario 1). This is supported by the high VAF of TET2 mutation compared with BRAF mutation in the bone marrow from patient A from Figure 3 in the present report. Alternately, other unknown molecular events at the level of HSCs or myeloid progenitors may promote BRAF and TET2 mutations (scenario 2). However, not all patients with CH develop myeloid neoplasms, as is evident in the current report (69% patients with CH had no concomitant myeloid malignancy). In this and other scenarios, TET2 mutant cells may support the outgrowth of BRAF mutant myeloid cells through secretion of cytokines, such as interleukin-6, which is almost universally elevated in ECD (scenario 3).^9,10  Conversely, MAPK-ERK–mutated cells could lead to TET2 mutations through their proinflammatory effects. The most common scenario, however, is when MAPK-ERK mutations lead to ECD without concomitant CH or other myeloid neoplasms (scenario 4).

Because the report by Cohen Aubart et al was a retrospective study of nonconsecutive ECD patients who underwent a bone marrow biopsy for various reasons, it is possible that the results are skewed toward higher CH mutations in their cohort. The lack of an age-matched control group further complicates the interpretation, especially given that the patients with TET2 mutations were older than others at diagnosis (mean age 67 vs 54 years). Therefore, the true prevalence of CH in ECD may not be higher than the age-matched population unless confirmed by other studies.

In previous reports, the development of a myeloid malignancy among ECD patients was invariably fatal.^2,4  Further research is needed to ascertain which ECD patients will develop myeloid neoplasms, including evaluation of the role of CH and bone marrow microenvironment. BRAF-inhibitor therapy in ECD has been reported to uncover underlying CMML due to paradoxical MAPK-ERK activation in non-BRAF–mutated myeloid cells, suggesting the need for some level of vigilance with rising monocyte counts among these patients.²  The implications of discovering CH are multifold (akin to monoclonal gammopathy of undetermined significance diagnosis), including increased patient anxiety, the need for appropriate counseling, and ongoing monitoring. Therefore, at this time it may be prudent to limit bone marrow biopsies with NGS testing to situations whereby peripheral blood abnormalities are present, as endorsed by the latest ECD guidelines.¹¹  Aside from myeloid neoplasms, CHIP also has significant associations with cardiovascular disease.⁵  Further evaluation of the cardiovascular outcomes of ECD patients with or without CH is needed, especially given that CH was associated with increased vascular involvement by ECD in the present report. These findings also raise the question whether patients with myeloid neoplasms and evidence of vasculitis or retroperitoneal infiltration have undiagnosed ECD. Although ECD is rare, several of these questions can be adequately answered through multi-institutional collaborations. Our understanding of ECD has improved exponentially in the last decade, leading to increased diagnosis, improved survival, and the first approval of a targeted therapy (vemurafenib). This highlights that rare disease research is feasible and fruitful, and the involvement of the next generation of investigators is critical to continue making advances in the field.