Does clonal hematopoiesis explain unexplained anemia?

In this issue of Blood, van Zeventer and her colleagues in The Netherlands report mutational profiling in 676 anemic patients ≥60 years old and a group of age-matched nonanemic controls.¹  These study subjects were selected from the LifeLines population cohort, which includes 22 108 people in that age band, most in the northern provinces of The Netherlands. The prevalence of anemia in the LifeLines cohort, just >3%, is considerably lower than in the US National Health and Nutrition Examination Survey (NHANES) III cohort, which may relate to better population health in The Netherlands compared with the United States or germline genetic differences.

Many older people are anemic, and it is not always clear why. In the third US NHANES III cohort, for example, >10% of community-dwelling Americans >60 years of age were found to be anemic when using World Health Organization (WHO) hemoglobin cutoffs to define anemia, including >20% of the oldest old (ie, those over age 85).²  About 2 out of every 3 NHANES anemia cases were due to nutritional deficiency, renal failure, or inflammation. The other one-third was considered “unexplained.”

In the 15 years since the NHANES anemia prevalence data were reported, investigators have proposed various hypotheses to explain unexplained anemia in the elderly as well other “idiopathic cytopenias of undetermined significance” (ICUS).³  These hypotheses include occult immune-mediated marrow suppression or premature blood cell destruction, stem cell “exhaustion,” or undiagnosed myelodysplastic syndromes (MDS).

Definitive diagnosis of immune-mediated cytopenias remains difficult (eg, immune thrombocytopenic purpura remains a diagnosis of exclusion), whereas hematopoietic cell exhaustion is a multifaceted and rather ill-defined phenotype. However, there is certainly suggestive evidence that MDS is underdiagnosed. In 1 Israeli hospital, for example, geriatric patients who were admitted to a ward for patients with cognitive impairment and who were noted to have minor blood count abnormalities underwent in-depth evaluation; 15% ultimately were proved to have MDS.⁴  Many elderly patients who might have MDS do not undergo full evaluation of mild cytopenias, especially very old patients with chronic health problems who are living in long-term care facilities in whom MDS, if diagnosed, would not be aggressively treated.

The high prevalence of anemia in the elderly is being reconsidered now that we know that somatic mutations are acquired in all tissues throughout the human lifespan, and that stable expanded blood cell populations derived from hematopoietic stem cells bearing acquired mutations that are associated with hematological neoplasia are present in almost everyone by middle age.^5,6  Most people with somatic mutations and clonal hematopoiesis have normal complete blood counts, however.⁷  This remains true even when the mutations are in genes associated with MDS or other myeloid neoplasms and are present at a variant allele frequency (VAF) ≥2%, which is near the detection or reporting threshold of common clinical next-generation sequencing assays and was used to define clonal hematopoiesis of indeterminate potential (CHIP).⁸ 

In the Dutch series, somatic mutations in blood cells at ≥1% were more frequent in anemic individuals (46.6%) than in controls (39.1%), which is also higher than the prevalence of CHIP in previous series. The relatively small difference in prevalence between anemic people and controls, however, suggests that the mutations do not account for the majority of anemias in elderly people, which is underscored by the fact that there was no difference between groups in the prevalence of the 3 most common CHIP mutations: DNMT3A, TET2, and ASXL1. Instead, other genes were more commonly mutated in anemic persons, including SF3B1, strongly associated with ring sideroblasts and ineffective erythropoiesis, and the dreaded TP53.

With follow-up, most clonal populations were stable over time and exhibited little change in VAF, at least during the length of the monitoring period. It is an unresolved question why this stability occurs so commonly. Mutations such as DNMT3A R882H give hematopoietic cells a growth advantage compared with wild-type cells, yet clonal sweeping with complete dominance of hematopoiesis is uncommon in the absence of secondary mutations, for unclear reasons.

Interestingly, given the relationship between CHIP and inflammation,⁹  mutations were more commonly detected in people thought to have anemia of inflammation compared with other anemia types, such as nutritional anemia. “Inflammation” is somewhat of a loosey-goosey concept, as there are many different inflammatory pathways and biomarkers. In the Dutch series, inflammation was defined by either elevated high-sensitivity C-reactive protein, unexplained leukocytosis, or an iron pattern consistent with inflammatory changes.

The Netherlands is the nation with the tallest people on the planet, and this new series underscores that size matters, at least when it comes to hematopoietic clones. Larger clones with a VAF >5% or multiple mutations were associated with inferior overall survival, but smaller clones were not. There is a growing body of evidence that larger clone size is associated with increased risk of AML development, cardiovascular disease, and all-cause mortality.

The clinical concept of “clonal cytopenias of undetermined significance” (CCUS; ie, ICUS with a demonstrated mutation) is important, but CCUS includes a heterogeneous patient group. Let us imagine a 74-year-old woman who comes to a hematology clinic with a hemoglobin count of 10 g/dL, mean cell volume of 100 fL, unremarkable white count and differential, platelet count of 150 × 10⁹/L, no obvious cause for her anemia such as B₁₂ or folate deficiency, and a nondiagnostic marrow aspirate. On next-generation sequencing, she is found to have a DNMT3A mutation at 4% VAF. Is that clone by itself enough to explain her cytopenias? Probably not, nor do clonal mutations explain anemia for the majority of elderly patients.

However, a 75-year-old man with similar blood and marrow findings who has 3 or 4 mutations found on sequencing, including both a splicing mutation and ASXL1 truncating variant with VAF >30%, effectively has “MDS without dysplasia.” The detected clonal process likely does explain his anemia, and the subset of multiple-mutant high-VAF CCUS is also associated with a substantial risk of progression to WHO-defined myeloid neoplasia.¹⁰ 

In the future, it will be helpful to have tests that can distinguish in the individual patient whether clonal hematopoiesis and cytopenias are both present yet unrelated, or are causal and connected. Knowing which clones are at greatest risk of causing subsequent clinical complications based on VAF, specific allele pattern, and other parameters (ie, which patterns are the most dangerous) is also an important goal, which will then allow us to design mitigation strategies for these complications. We are a few skate lengths closer to achieving those goals, thanks to this new series (see figure).