Clonal hematopoiesis – mutation-specific adaptation to environmental change

Abstract

Clonal hematopoiesis of indeterminate potential (CHIP) describes a widespread expansion of genetically variant hematopoietic cells that increases exponentially with age and is associated with increased risks of cancers, cardiovascular disease, and other maladies. Here we discuss how environmental contexts associated with CHIP, such as old age, infections, chemotherapy, or cigarette smoking, alter tissue microenvironments to facilitate the selection and expansion of specific CHIP mutant clones. Further, we consider major remaining gaps in knowledge, including intrinsic effects, clone size thresholds and factors affecting clonal competition, that will determine future application of this field in transplant and preventive medicine.

Introduction

Hematopoietic stem cells (HSCs) reside in the bone marrow where they divide, differentiate, and self-renew to maintain all blood cells throughout an organism’s life. (Pinho and Frenette, 2019) Over time, HSCs naturally acquire somatic mutations, some of which enable selective advantages over other HSCs in a context-dependent fashion (Figure 1). As a result of this competition, oligoclonality in the bone marrow has been observed, initially based on patterns of X chromosome inactivation and later confirmed in normal individuals harboring TET2 mutations. (Busque et al., 2012) The term clonal hematopoiesis of indeterminate potential (CHIP) was first coined in 2015 by Steensma et al. to define a phenotype in which hematopoietic cells harboring somatic mutations clonally expand in the absence of hematological disease. (Steensma et al., 2015) A number of studies have defined the genetic landscape of CHIP, with most mutations occurring in pre-leukemic driver genes that are recurrently implicated in the pathogenesis of hematological malignancies and disorders.(Bolton et al., 2020, Genovese et al., 2014, Miller and Steensma, 2020, Jan et al., 2017) While CHIP was originally defined by detection of mutations in genes related to hematological neoplasms at a variant allele frequency (VAF) of ≥ 2%, newer methods now enable detection of CHIP at a much lower VAF. (Steensma et al., 2015)

Figure 1: Environmental stressors promote mutation-specific clonal expansion.

Over an organism’s lifetime, HSCs encounter cellular stresses which may induce somatic mutations in common driver genes. Depending on the context, somatic mutations may confer a selection advantage for certain clones to persist and expand. Environmental effects on clonal competition are mutation specific. Genotoxic stressors such as chemotherapy and radiation provide a selection advantage for clones bearing mutations in p53, PPM1D, and CHEK2, whereas infection and inflammation promote the expansion of clones with mutations in TET2 and DNMT3A. Studies on the metabolic effects of clonal expansion suggest that metabolites, like vitamin C, may impact the function of mutations, such as TET2.

CHIP is found, by conservative measures, in more than 10% of individuals by the eighth decade of life and poses increased risks of adverse outcomes including severe infection, kidney disease, hematological and non-hematological cancers, cardiovascular disease, and all-cause mortality. (Jaiswal et al., 2017, Genovese et al., 2014, Zink et al., 2017, Jaiswal et al., 2014, Kessler et al., 2022, Kar et al., 2022, Bolton et al., 2021, Zekavat et al., 2021a, Dawoud et al., 2020, van Deuren et al., 2021) Despite the increased prevalence of CHIP with age, CHIP mutations appear to be relatively stable over time in the absence of additional stressors. (Midic et al., 2020) This stability implies that extrinsic factors provide opportunities for somatic mutations to undergo “somatic evolution,” facilitating the selection and expansion of clones in specific microenvironments.(Laconi et al., 2020) There is an increasing body of evidence showing that particular stressors can drive expansion of specific CHIP mutant clones (Figure 1). Studies exploring the mechanisms by which specific mutant clones outcompete other HSCs and expand in certain contexts have provided insight into the selective environments that drive their expansion. Understanding the differential fitness landscapes that promote the selective expansion of CHIP variants will enhance our ability to not only manage clonal expansion but also reduce risks associated with CHIP. (King et al., 2020)

CHIP and Disease Risk

CHIP is associated with a widening array of diseases, many of which are age-associated (Table 1). Patients with CHIP have an increased risk of developing hematological malignancies, with the effect size varying according to VAF, clonal complexity, mutation type, and gene(s) mutated.(Desai et al., 2018, Abelson et al., 2018, Zink et al., 2017, Genovese et al., 2014, Jaiswal et al., 2014) Deep sequencing analysis has revealed that the risk of individuals with CHIP developing acute myeloid leukemia (AML) may occur at VAFs as low as 0.005-0.01,(Young et al., 2019, Desai et al., 2018, Abelson et al., 2018) with further increase in AML risk proportionate to VAF.(Abelson et al., 2018, Genovese et al., 2014) Mutations commonly identified in CHIP include those in epigenetic modifiers (TET2, DNMT3A, IDH1, IDH2, and ASXL1), splicing factors (U2AF1, SRSF2), DNA damage repair genes (PPM1D, TP53) and inflammatory mediators (JAK2, STAT3, MYD88), most of which are known drivers of hematological cancers. The risk of future blood cancers and overall decreased survival varies by mutation, with mutations in DNMT3A and TET2 conferring a lesser risk of progression compared to PPM1D, TP53.(Genovese et al., 2014, Abelson et al., 2018, Chou et al., 2011, Gelsi-Boyer et al., 2012, Park et al., 2020, Yuan et al., 2016a, Ley et al., 2010, Jaiswal et al., 2014, Desai et al., 2018)

Table 1:

Suspected CHIP-mediated diseases by driver mutation

Gene	Function(s) in HSCs	Suspected CHIP-associated outcomes
DNA Damage Response
PPM1D	• Protein phosphatase Mn2+/Mg2+ dependent 1D; negative regulator of DDR pathway	• t-MN (Hsu et al., 2018)   • t-MDS (Lindsley et al., 2017)   • Heart failure (Yura et al., 2021)   • CKD (Dawoud et al., 2021)   • MDS (Papaemmanuil et al., 2013, Haferlach et al., 2014)   • t-MN (Bolton et al., 2020, Wong et al., 2015)   • Heart failure (Sano et al., 2021)   • t-MDS(Lindsley et al., 2017, Wong et al., 2015)   • Atherosclerosis(Zekavat et al., 2021b)   • MDS (Janiszewska et al., 2018)
TP53	• Regulator of cellular proliferation and differentiation   • Transcriptional factor regulating cellular stressing cellular stress   • Epigenetic modifier
Chek2	• Checkpoint Kinase 2; regulator of DDR
Epigenetic Modifiers
DNMT3a	• De novo DNA methyltransferase	• MDS(Haferlach et al., 2014, Papaemmanuil et al., 2013)   • MPN(Lundberg et al., 2014)   • Atherosclerosis- (Rauch et al., 2018)   • Renal fibrosis (Sano et al., 2018a)   • HF/CVD/MI(Bick et al., 2020a, Abplanalp et al., 2021, Jaiswal et al., 2017, Sano et al., 2018a, Dorsheimer et al., 2019)   • Stroke (Bhattacharya et al., 2021, Jaiswal et al., 2014)   • AML (Young et al., 2019)   • Osteoporosis (Kim et al., 2021)   • Pre-mature menopause (Honigberg et al., 2021)   • Liver-GVHD (Newell et al., 2021)   • Aortic valve stenosis (Mas-Peiro et al., 2020)   • Acute-GVHD SCT (Oran et al., 2021)
TET2	• DNA methylcytosine dioxygenase	• COPD(Buscarlet et al., 2017, Zink et al., 2017, Miller et al., 2021a)   • Stroke (Bhattacharya et al., 2021, Jaiswal et al., 2014)   • Gout (Agrawal et al., 2021)   • Lung cancer progression (Nguyen et al., 2021)   • MDS (Haferlach et al., 2014, Papaemmanuil et al., 2013)   • Heart failure/CVD/MI (Sano et al., 2018b, Wang et al., 2020, Bick et al., 2020a, Jaiswal et al., 2017, Dorsheimer et al., 2019, Yu et al., 2021)   • Renal fibrosis (Sano et al., 2018a)   • Atherosclerosis (Fuster et al., 2017, Rauch et al., 2018)   • MPN(Lundberg et al., 2014)   • CKD (Dawoud et al., 2021)   • MDS(Haferlach et al., 2014)
ASXL1	• Polycomb-associated protein involved in epigenetic regulation and transcription	• MPN(Lundberg et al., 2014, Nangalia et al., 2013, Vannucchi et al., 2013)   • HF/CVD/MI (Jaiswal et al., 2014, Jaiswal et al., 2017, Yu et al., 2021)   • Stroke (Jaiswal et al., 2014)
IDH1/IDH2	• Isocitrate dehydrogenase 1/2; metabolizes Isocitrate to α-ketoglutarate	• MPN(Lundberg et al., 2014, Vannucchi et al., 2013)
Cellular signaling
JAK2	• Non-receptor tyrosine kinase; signals through JAK/STAT pathway	• MPN(Wolach et al., 2018, Lundberg et al., 2014)   • Thrombosis (Wolach et al., 2018)   • Atherosclerosis (Wang et al., 2018)   • CVD(Jaiswal et al., 2017)   • Heart failure (Sano et al., 2019, Yu et al., 2021)   • Aortic aneurysms (Yokokawa et al., 2021)   • Pulmonary HTN (Kimishima et al., 2021)   • CKD (Dawoud et al., 2021)
Spliceosome
SF3B1	• RNA splicing	• MDS(Haferlach et al., 2014, Papaemmanuil et al., 2013)
SRSF2	• RNA splicing	• MDS(Haferlach et al., 2014, Papaemmanuil et al., 2013)   • MPN (Vannucchi et al., 2013)
U2AF1	• RNA splicing	• MDS(Haferlach et al., 2014, Papaemmanuil et al., 2013)

DNMT3A and TET2 are the most commonly mutated genes in CHIP. (Buscarlet et al., 2017, Feusier et al., 2021, Genovese et al., 2014, Jaiswal et al., 2014, Xie et al., 2014) DNMT3A mutations are present in at least 20% of AML patients, (Gaidzik et al., 2013, Lauber et al., 2020, Park et al., 2020, Yan et al., 2011, Ley et al., 2010) with roughly 60% of these occurring at the R882 codon. (Gaidzik et al., 2013, Lauber et al., 2020, Ley et al., 2010, Park et al., 2020, Ribeiro et al., 2012) Mutations at this hotspot locus, located within the methyltransferase catalytic domain, have been linked with a higher risk of AML development and poorer outcomes compared to other mutations within the same gene. (Gaidzik et al., 2013, Kumar et al., 2018, Lauber et al., 2020, Ley et al., 2010, Renneville et al., 2012, Ribeiro et al., 2012, Yuan et al., 2016a, Young et al., 2019) Similarly, mutations in TET2, a methylcytosine dioxygenase, occur at high frequencies in myeloid neoplasms (Abdel-Wahab et al., 2009, Delhommeau et al., 2009, Xie et al., 2014), and patients harboring mutations in TET2 tend to have decreased survival and response to therapy. (Wang et al., 2019) DNMT3A and TET2 regulate methylation and demethylation of the genome, respectively, though through different mechanisms.(Izzo et al., 2020) Despite these mechanistic differences, functional studies suggest that mutations in both DNMT3A (Lauber et al., 2020, Park et al., 2020, Russler-Germain et al., 2014, Sandoval et al., 2019, Yan et al., 2011) and TET2 (Lopez-Moyado et al., 2019, Rasmussen et al., 2015, Cimmino et al., 2017) alter the expression of genes involved in proliferation, differentiation, and oncogenesis. These epigenetic changes allow the mutant cells to propagate and persist, enabling the acquisition of other mutations that further drive leukemogenesis. (Bezerra et al., 2020, Celik et al., 2015, Guryanova et al., 2016, Kronke et al., 2013, Loberg et al., 2019, McKerrell et al., 2015, Meyer et al., 2016, Yang et al., 2016, Jan et al., 2017) While the mechanistic roles of DNMT3A and TET2 in malignant hematopoiesis have been thoroughly summarized elsewhere, (Chaudry and Chevassut, 2017, Im et al., 2014, Venugopal et al., 2021, Yang et al., 2015, Bowman and Levine, 2017, Nakajima and Kunimoto, 2014, Feng et al., 2019) it is worth noting that blood cells carrying CHIP mutations may not only increase the risk of AML but also may impact non-hematological tumors by modulating the immune microenvironment.(Kleppe et al., 2015) Indeed, CHIP has been associated with a variety of solid tumors in multiple large cohort studies,(Bolton et al., 2020, Kar et al., 2022, Kessler et al., 2022) strongly suggesting that CHIP may also affect the progression of solid tumors.

In addition to its impact on cancer development and prognosis, CHIP has been noted to affect autologous stem-cell transplant (ASCT) outcomes for lymphoma. Patients with CHIP mutations had higher incidence of therapy-related myeloid neoplasms (TMNs) following ASCT. Furthermore, within this cohort, patients harboring PPM1D mutations had overall lower survival.(Gibson et al., 2017b) An increased prevalence of unexplained cytopenias in patients receiving ASCT of hematopoietic cells harboring CHIP mutations has been reported, suggesting graft dysfunction. (Gibson et al., 2017a) However, others report no adverse outcomes with CHIP post-ASCT, and even improved survival, suggesting that adverse effects may be mutation-specific (Grimm et al., 2019, Heini et al., 2021).

CHIP also has recently been implicated in impacting both risk and response to infections. Patients harboring CHIP were found to have increased risk of severe COVID-19, sepsis and other infections.(Bolton et al., 2021, Dawoud et al., 2020, Jaiswal and Ebert, 2019, Kessler et al., 2022, Zekavat et al., 2021a) These risks may be attributable to impaired function of downstream immune cells, which has been reported for cells bearing CHIP mutations.(Jaiswal and Ebert, 2019, Zekavat et al., 2021a)

One phenomenon of great interest related to CHIP is its link to cardiovascular disease and all cause-mortality.(Jaiswal et al., 2014, Sano et al., 2018c, Libby et al., 2019, Dorsheimer et al., 2019, Cremer et al., 2020, Abegunde et al., 2018, Kar et al., 2022, Kessler et al., 2022) One of the earliest studies found that CHIP carriers had a two-fold risk of coronary heart disease and ischemic stroke (Jaiswal 2014). Subsequent studies in patient cohorts found a four-times greater risk of having early myocardial infarction (MI),(Jaiswal et al., 2017) and congestive heart failure in CHIP patients(Dorsheimer et al., 2019), independent of traditional cardiovascular risk factors. Recently, however, two recent large epidemiological studies using the UK Biobank indicate that the increased risks of cardiovascular disease may be less dramatic (HR 1.1, 1.03-1.17).(Kar et al., 2022, Kessler et al., 2022) Specific mutations, including TET2, DNMT3A and ASXL1 showed an increased incidence of coronary artery disease, coronary artery calcifications, and early MI, with risks of cardiovascular disease that correlate with VAF.(Jaiswal et al., 2014, Bick et al., 2020a, Kar et al., 2022, Jaiswal et al., 2017) Studies using both human samples and mouse models revealed the pro-inflammatory environment generated by the mutant clones drives adverse cardiovascular outcomes.(Jaiswal et al., 2017, Fuster et al., 2017, Sano et al., 2018a, Bick et al., 2020a) However, differing pro-inflammatory phenotypes drive cardiovascular dysfunction and remodeling in CHIP with mutations in PPM1D (Yura et al., 2021), TP53 (Sano et al., 2021), DNMT3A (Abplanalp et al., 2021, Heyde et al., 2021, Sano et al., 2018c) and TET2 (Sano et al., 2018b, Wang et al., 2020), respectively. CHIP also has been linked to COPD,(Buscarlet et al., 2017, Miller et al., 2021a, Zink et al., 2017) diabetes,(Fuster et al., 2020, Jaiswal et al., 2014) psychiatric illnesses,(Zink et al., 2017) early menopause,(Honigberg et al., 2021) osteoporosis,(Kim et al., 2021) liver disease,(Wong et al., 2022) renal dysfunction,(Dawoud et al., 2021) and epigenetic aging.(Robertson et al., 2019) The correlation between these diseases and CHIP could be manifestations of tissue dysfunction due to common causes, like aging, smoking, and inflammation; however, it is possible that CHIP-driven inflammation contributes to the development of these diseases. As we discuss below, inflammation itself is a driver of CHIP, setting up a feed-forward loop of CHIP and disease. Thus, CHIP has important implications for the future management of patients, most immediately in relation to cardiovascular outcomes and precision oncology.(Libby et al., 2019, Miller and Steensma, 2020, Steensma, 2018, Sidlow et al., 2020)

It is worth noting that the factors that drive CHIP may have little to do with the mechanisms by which mutant clones promote disease. Epidemiologic studies will continue to be important to uncover key disease associations. Importantly, understanding what drives CHIP and its health consequences will provide two different avenues for preventive medicine and therapeutic management, respectively. With an eye toward prevention, here we will highlight known associations between environmental stressors and CHIP, including mechanisms of clonal expansion where known.

Chemotherapy and Radiation as Drivers of CHIP

Chemotherapies and radiation destroy malignant cells by inducing DNA damage, impairing DNA replication, damaging DNA repair mechanisms, inhibiting pro-survival signaling pathways, impairing transcription and translation, and evoking metabolic and cellular stress.(Krisl and Doan, 2017) In addition to cancer treatments, chemotherapy and radiation are useful in managing bone marrow diseases,(Lee et al., 2018) transplants,(Krisl and Doan, 2017) immunodeficiencies,(Lum et al., 2019) and autoimmune diseases(2004) where cytotoxic conditioning regimens make space for engraftment, prevent rejection, and dampen an overactive immune system. These reagents universally impair the hematopoietic system either as the primary goal or as an off-target effect of tumor targeting.(Shao et al., 2013) Quiescent HSCs are more protected against the acute stress of chemotherapy and radiation in part via induction of a strong P53-dependent DNA damage response (DDR), which promotes senescence instead of cell death(Mohrin et al., 2010). However, HSCs eventually lose clonogenic activity, undergo cell death, and have decreased fitness, defined as overall ability to persist and expand, in a P53-dependent manner.(Marusyk et al., 2010, Bondar and Medzhitov, 2010) Furthermore, the quiescent nature of HSCs limits DNA repair primarily due to the use of error-prone non-homologous end joining.(Mohrin et al., 2010) Thus, HSC survival comes at the expense of increased risk of mutagenesis and dampened HSC expansion.(Mohrin et al., 2010, Schoedel et al., 2016) Specific pre-existing CHIP mutants are evolutionarily advantaged to expand and persist under the stress of chemotherapy and radiation. Multiple reports have shown patients with a prior history of cancer and cancer treatment are more likely to have certain CHIP mutations (Bolton et al., 2020, Boucai et al., 2018, Coombs et al., 2017, Olszewski et al., 2019, Hsu et al., 2018, Kahn et al., 2018, Wong et al., 2018), with selective clonal expansion dependent on the mechanism of the cytotoxic stressor as described below (see also Table 2).

Table 2:

Extrinsic Mechanisms of Clonal Expansion

Class	Gene	Mechanism	Description	Model	Reference(s)
Aging
	CHIP	next generation sequencing	Cross sectional study association	human	Jaiswal et al., 2014(Jaiswal et al., 2014), Xie et al., 2014(Xie et al., 2014), Buscarlet et al., 2017(Buscarlet et al., 2017), van Zeventer et al., 2021(van Zeventer et al., 2021)
	DNMT3A, TET2	exome sequencing	Association Study between CH risk and longer leukocyte telomere length	human n=200,453	Kar et al., 2022 (Kar et al., 2022)
	ASXL1	Truncated ASXL1	Truncated ASXL1 expands in murine model of aging via interactions with BAP1 to activate mTORC for survival	mouse	Fujino et al., 2021 (Fujino et al., 2021)
Chemotherapy/Radiation
	CHIP	Radioactive iodine (RAI)	Cross-sectional study showed RAI was linked with increased prevalence of CH	human n= 279	Boucai et al., 2018 (Boucai et al., 2018)
	CHIP	Chemotherapy followed by ASCT	Longitudinal association study found MRD-samples treated with chemotherapy-ASCT was linked to CHIP expansion	human n=149	Eskelund et al., 2020 (Eskelund et al., 2020)
	PPM1D	Chemotherapy (including topoisomerase I and II inhibitors, taxanes, platinum-based therapies, etoposide cytarabine, doxorubicin), XRT, Radionuclide	Multiple cross-sectional studies, mouse model, and in vitro experiments linked PPM1D to specific cytotoxic stressors	human 10,138 human n=5,649 human n=119 human n=686 human n=1,185 mouse,	Bolton et al., 2020 (Bolton et al., 2020) Coombs et al., 2017 (Coombs et al., 2017) Wong et al., 2018 (Wong et al., 2018) Swisher et al., 2016 (Swisher et al., 2016) Kim et al., 2019 (Kim et al., 2019) Hsu et al., 2018 (Hsu et al., 2018) Kahn et al., 2018 (Kahn et al., 2018)
	CHEK2	Chemotherapy (Topoisomerase II inhibitors, Platinum based therapies)	Cross-sectional study	human n= 10,138	Bolton et al., 2020 (Bolton et al., 2020)
	TP53	Chemotherapy (including platinum base therapies, taxanes, 5-FU), XRT	Multiple cross-sectional studies, and mouse models that show TP53 selective advantage under these cytotoxic stressors	human n= 10,138 human n= 5,649 human n=119 Mouse models	Bolton et al., 2020 (Bolton et al., 2020) Coombs et al., 2017(Coombs et al., 2017) Wong et al., 2018 (Wong et al., 2018) Bondar et al., 2010(Bondar and Medzhitov, 2010) Marusyk et al., 2010(Marusyk et al., 2010) Wlodarski et al., 1998 (Wlodarski et al., 1998)
	TP53	n-ethyl-n-nitrosourea	Treatment of Tp53+/− HSPCs showed competitive advantage	mouse	Wong et al., 2015 (Wong et al., 2015)
	TET2	Chemotherapy (including doxorubicin and cytosine arabinoside)	TET2KOAML cells are less sensitive to treatment; cross sectional study linked to TET2.	AML cell lines; human n=5,649	Morinishi et al., 2020 (Morinishi et al., 2020) Coombs et al., 2017 (Coombs et al., 2017)
	TET2	Spaceflight (XRT)	Astronauts that experienced space flight had earlier expansion of TET2.	human, n=2	Mencia-Trinchant et al., 2020 (Mencia-Trinchant et al., 2021)
	DNMT3A	Busulfan	Dnmt3a−/− HSCs expand with treatment.	mouse	Chen et al., 2020 (Chen et al., 2020)
Tobacco/Smoking
	ASXL1	Current and former smoking history	Multiple cross-sectional studies	human n=5,649 n=502,524 n= 200,453 n= 628.388 n= 10,138	Coombs et al., 2017 (Coombs et al., 2017) Dawoud et al., 2020(Dawoud et al., 2020) Kar et al., 2022 (Kar et al., 2022) Kessler et al., 2022(Kessler et al., 2022) Bolton et al., 2020 (Bolton et al., 2020)
	DNMT3A, TET2	Current/former tobacco use and smoking	Cross-sectional study	human n=5,649 n=502,524	Coombs et al., 2017 (Coombs et al., 2017) Dawoud et al., 2020(Dawoud et al., 2020)
	PPM1D	Current/former tobacco use	Cross-sectional study	human n=5,649	Coombs et al., 2017 (Coombs et al., 2017
	DNMT3A, SRSF2, SR3B1	Smoking status (current)	Cross-sectional study	human n=200,453	Kar et al., 2022 (Kar et al., 2022)
	CHIP	Smoking history	Cross-sectional study	human n=12,380	Genovese et al., 2014 (Genovese et al., 2014)
Inflammation
	DNMT3A	Elevated serum MIP1a	Cross sectional study showing increased risk of DNMT3A-CHIP	human n=200,453	Kar et al., 2022 (Kar et al., 2022)
	DNMT3A	Chronic IFNy signaling via M. avium infection	Infection promotes DNMT3A KO expansion due to resistance to inflammation-mediated myeloid differentiation.	mouse	Hormaechea-Agulla et al., 2021 (Hormaechea-Agulla et al., 2021)
	TET2	LPS injection	TET2 clones selectively expanded via lncRNA Morrbid and aTLR4/NK-kB/IL-6/STAT3 pathway	mouse	Cai et al., 2018 (Cai et al., 2018)
	TET2	TNFa signaling	TET2 selectively expanded with direct TNFa signaling	mouse	Abegunde et al., 2018 Abegunde et al., 2018)
	TET2	Disruption of intestinal barrier integrity	Increased microbial translocation and signaling, particularly from Lactobacillus, promotes inflammation and TET2 expansion	mouse	Meisel et al., 2018 (Meisel et al., 2018)
	TET2	broad spectrum antibiotics	Antibiotics slowed the expansion of TET2-deficient clones	mouse	Zeng et al., 2019 (Zeng et al., 2019)
	CEBPa	Direct injection of IL-1b	IL-1b promotes KO CEBPa MPP3 clonal expansion, while CEBPa activates wildtype competitors for myeloid expansion and exhaustion.	mouse	Higa et al., 2021 (Higa et al., 2021)
	ASXL1	Tissue editing With Inducible Stem cell Tagging via Recombination (TWISTR) knock-in of CH genes and single cell analysis	Single cell sequencing of knocked-in CHIP homologs zebrafish HSCs demonstrated resistance to inflammation in CH stem cells and promotion of cytokine production in myeloid lineage cells.	zebrafish	Avagyan et al., 2021(Avagyan et al., 2021)
	TP53	poly:IC serial injection	TP53 clones were significantly expanded via IFNy production caused by poly:IC treatment in competitive transplant models	mouse	Rodriguez-Meira et al., 2022 (Rodriguez-Meira et al., 2022)
Autoimmunity
	DNMT3A and PPM1D	Inflammatory environment caused by inflammatory bowel disease	Cross sectional study found higher prevalence of CHIP in patients with ulcerative colitis	human n=187	Zhang et al., 2019 (Zhang et al., 2019)
HSCT
	DNMT3A	Autologous HSCT	Cross sectional study found DNMT3A associated with HSCT	human n= 35	Wong et al., 2018 (Wong et al., 2018)
	DNMT3A	Murine competitive transplant	DNMT3A null HSCs outcompete wildtype HSCs in murine transplant settings	mouse	Challen et al., 2011(Challen et al., 2011) Jeong et al., 2018(Jeong et al., 2018)
	PPM1D	Murine competitive transplant	Truncated PPM1D R451X HSCs fails to outcompete wildtype HSCs in competitive murine transplant settings	mouse	Hsu et al., 2018(Hsu et al., 2018)
	TET2	Serial HSC transplant	Tet2 null HSCs exhaust similarly to wildtype HSCs in serial transplant settings	mouse	Ostrander et al., 2020(Ostrander et al., 2020)
	ASXL1	Murine competitive transplant	Truncated ASXL1 are at a disadvantage in engraftment and cell function in a murine transplant setting	mouse	Fujino et al., 2021(Fujino et al., 2021)
	SRSF2	Murine competitive transplant	Heterozygous Srsf2 P95H HSCs cannot outcompete WT HSCs in engraftment or lineage output in competitive murine transplants	mouse	Kon et al., 2018(Kon et al., 2018)
	SF3B1	Murine competitive transplant	Sf3b1-K700E mutant HSCs are at a disadvantage compared to wild-type HSCs in both young or old recipient mice.	mouse	Mupo et al., 2017(Mupo et al., 2017)
	U2AF1	Murine competitive transplant	U2AF1 chimerism in transplanted mice is significantly reduced in mutant U2AF1 variants: S34F, Q157P, and Q157R.	mouse	Yoshida et al., 2011(Yoshida et al., 2011)
Metabolic
	TET2	Hyperglycemia	Hyperglycemia drives TET2 myelopoiesis and MPN.	mouse	Cai et al., 2021(Cai et al., 2021)
	TET2	Vitamin C deficiency	Cross-sectional study where lower plasma vitamin C concentrations was associated with a higher frequency of TET2 mutations	human n = 215	Chen et al., 2021(Chen et al., 2021)
	TET2	Apolipoprotein B	High ApoB was associated with TET2 clonal expansion, whereas alcohol consumption was protective in this cross-sectional study	human n = 200,453	Kar et al., 2022(Kar et al., 2022)
	TET2, DNMT3A, ASXL1	BMI	High BMI was protective against DNMT3A clonal expansion, but associated with TET2 and ASXL1 CHIP in this cross-sectional study	human n = 628,388	Kessler et al., 2022(Kessler et al., 2022)
	DNMT3A	Fatty bone marrow model	Self-renewal and engraftment advantage was observed in DNMT3A mutant cells and was mediated by IL-6	mouse	Zioni et al., 2022(Zioni et al., 2021)
	CHIP	Unhealthy diet, low LDL, Low HDL, insulin resistance, and hypertension are associated with increased risk of CHIP.	Several cross sectional or longitudinal association studies link CHIP with metabolic syndrome.	human n = 44,111	Bhattacharya et al., 2021(Bhattacharya et al., 2021), Kar et al., 2022(Kar et al., 2022), van Deuren et al., 2022 (van Deuren et al., 2021)
Other
	CHIP	Down syndrome	Observational study found early signs of CHIP in Down’s syndrome patients	human n=51	Liggett et al., 2021 (Liggett et al., 2021)
	DNTM3A, PPM1D	Growth factor usage, RBC transfusion	Cross sectional study-linked association	human n=5,649	Coombs et al., 2017 (Coombs et al., 2017)
	TET2	Atherosclerosis	Atherosclerosis accelerates HSC proliferation and expansion of TET2−/− cells	Mathematical modeling, mouse	Heyde et al., 2021 (Heyde et al., 2021)

DDR genes

Mutations in DDR genes (TP53, PPM1D, and CHEK2) have the strongest association with prior cancer treatment, and clones with mutations in these genes expand more under the stress of specific cytotoxic therapies.(Bolton et al., 2020, Coombs et al., 2017, Swisher et al., 2016, Wong et al., 2015) Mutations in DDR genes are infrequent in the absence of prior cytotoxic exposure,(Xie et al., 2014, Genovese et al., 2014, Jaiswal et al., 2014, Eskelund et al., 2020) but prevalent in patients with TMN, indicating the selective advantage under cytotoxic stress.(Bolton et al., 2020, Gibson et al., 2017b, Hsu et al., 2018, Lindsley et al., 2017)

Studies have shown that PPM1D mutations often occur in exon 6 and confer resistance to chemotherapeutics, especially platinum-based therapies.(Bolton et al., 2020, Hsu et al., 2018, Kahn et al., 2018, Kim et al., 2019) PPM1D mutations are typically gain of function truncation mutations that stabilize the protein and suppress the DDR, enabling mutant clones to resist apoptosis.(Hsu et al., 2018, Kahn et al., 2018) Similar to the selective advantages observed in large patient cohorts,(Bolton et al., 2020) in vitro studies show that PPM1D mutants are resistant to specific cytotoxic therapies (cisplatin, etoposide, doxorubicin, cytarabine) but not others (5-fluorouracil (5-FU), vincristine).(Hsu et al., 2018, Kahn et al., 2018) These differences are likely explained by these treatments’ differing mechanisms of action. For instance, vincristine impairs mitosis by microtubule interference and inhibits trafficking of DNA repair proteins, including P53, which interacts with PPM1D.(Kahn et al., 2018, Poruchynsky et al., 2015) It is plausible that microtubule interference blocks PPM1D from interacting with downstream DDR mediators, thereby preventing its ability to inhibit apoptosis.(Kahn et al., 2018, Poruchynsky et al., 2015)

Mutations in TP53 are a strong risk factor in the development of TMN.(Abelson et al., 2018, Desai et al., 2018, Gillis et al., 2017, Wong et al., 2015) P53 has multiple functions and may regulate HSC response to cytotoxic stress by regulating apoptosis,(Wu et al., 2005) proliferation,(Bondar and Medzhitov, 2010, Wlodarski et al., 1998, Wu et al., 2005) quiescence and self-renewal.(Chen et al., 2008, Liu et al., 2009) Cytotoxic drivers of TP53 mutant expansion include platinum, radiation, and taxane-based therapies.(Bolton et al., 2020) Furthermore, in vitro studies suggest that 5-FU(Wlodarski et al., 1998) and doxorubicin(Sano et al., 2021) confer a selective advantage for TP53 mutant clones. The expansion of TP53 mutant clones may be dependent on mutation type and may occur via multiple mechanisms, as TP53 mutations have been suggested to cause both gain- and loss-of-function. Interestingly, TP53 mutation has been reported to promote DNA H3K27 trimethylation via enhanced interaction with EZH2, thereby reinforcing HSC-self renewal during radiation stress. (Chen et al., 2019) Whether this mechanism underlies selective CHIP expansion of TP53 mutants during other cytotoxic stresses remains unknown, though it is likely that both DDR impairment and epigenetic changes underlie the selective advantages of this mutant.

Epigenetic modifiers

The role of cytotoxic stress on clones bearing mutations in epigenetic modifiers is less clear. A few small studies have shown clonal expansion in non-leukemic HSC clones harboring mutations in the epigenetic modifiers DNMT3A and TET2.(Wong et al., 2016) Coombs et al showed cytotoxic therapy was associated with TET2,(Coombs et al., 2017) but this was not recapitulated in later large cohort studies.(Bolton et al., 2020) Conversely, mutations in DNMT3A and TET2 are known to confer treatment resistance to chemotherapies in AML, and numerous studies have linked specific DNMT3A^mut in AML with poor clinical outcomes.(Lauber et al., 2020, Ley et al., 2010, Ribeiro et al., 2012, Yuan et al., 2016a) For example, DNMT3A^R882mut protects against anthracycline-mediated cell death in AML via an impaired chromatin remodeling response that prevents an appropriate DDR.(Guryanova et al., 2016) Restoration of this pathway sensitized DNMT3A^R882mut cells to anthracycline.(Guryanova et al., 2016) Conversely, administration of azacytidine, a DNA methyltransferase inhibitor, suppressed Dnmt3A^R882mut clones and prolonged survival in AML patients harboring DNMT3A^R882mut. (Scheller et al., 2021) In mouse models, DNMT3A^null cells expand following exposure to busulfan, but not 5-FU, highlighting the specificity of certain stressors to drive clonal expansion.(Chen et al., 2020) The effects of chemotherapies on DNMT3A^mut clonal selection may also be specific to the type of mutation.(Yuan et al., 2016b)

For TET2, one proposed mechanism driving treatment resistance is that hypermethylation caused by TET2 deletion promotes a quiescent stem-cell like state, desensitizing cells to chemotherapies such as cytosine arabinoside and doxorubicin.(Morinishi et al., 2020) The use of hypomethylating agents to target specific TET2^mut may prove beneficial in sensitizing TET2^mut cells to chemotherapy.(Stölzel et al., 2021)

The impact of mutations in IDH1/2 and ASXL1 on chemoresistance in AML remain unclear.(Brunner et al., 2019, Molenaar et al., 2018, Ni et al., 2020, Paschka et al., 2010, Paschka et al., 2015) Determining the mechanisms by which chemotherapy interacts with DNMT3A and other epigenetic modifier mutant clones will provide greater insights on how these mutant clones respond to cytotoxic pressures. (Chaudry and Chevassut, 2017, Metzeler et al., 2012, Venugopal et al., 2021, Yang et al., 2015, Yang et al., 2021)

Spliceosome complex proteins

Clones with mutations in spliceosome proteins (SRSF2 SF3B1, and U2AF1) are more likely to persist in AML patients following chemotherapy and are observed in a high proportion of secondary AMLs.(Lachowiez et al., 2021) Still, whether spliceosome mutants have selective advantages with certain chemotherapeutics or whether selection is a consequence of other genetic lesions within the leukemic clones remains unknown.

Finally, the clinical correlation between the selective advantage of non-leukemic epigenetic and spliceosome driver mutations in CHIP under cytotoxic therapy may be limited by the type of cytotoxic therapies used in large cancer patient cohort studies.(Bolton et al., 2020, Coombs et al., 2017) While prior studies in leukemic patients suggest epigenetic modifiers and spliceosome regulators play important roles in response to therapy, how various cytotoxic therapies regulate non-leukemic CHIP clones remains unclear. Further clinical and mechanistic studies are needed to determine whether cytotoxic drivers negatively or positively impact these classes of CHIP mutants. Expanding our understanding of these selective processes could revolutionize personalized medicine, such that treatment regimens are carefully chosen to avoid the selection of certain mutant clones.

Immunotherapies

Current cancer research has focused on the development of immunotherapies to prime, augment, or sustain immune responses to cancer. These therapies pose a greater risk of immune-mediated stress compared to the cytotoxic therapies described above.(Esfahani et al., 2020) Immune checkpoint inhibitors (ICIs) may induce autoimmunity, including immune-mediated destruction of the hematological compartment,(Davis et al., 2019) while chimeric antigen receptor-T cell (CAR T) therapy is known to induce cytokine release syndrome mediated largely by interleukin 1 (IL-1) and IL-6.(Norelli et al., 2018) To date, the impact of immunotherapies on CHIP remains controversial, with some studies indicating that immunotherapies may promote expansion of CHIP clones but others reporting no impact on clonal expansion. (Bolton et al., 2020, Miller et al., 2020)

While there is no consensus about the impact of immunotherapy on CHIP emergence, CHIP mutations are known to affect CAR T cell function. A case study found CAR T cells harboring a TET2 CHIP mutation had enhanced expansion, prolonged activation, increased cytokine production, and memory-like properties,(Fraietta et al., 2018) consistent with a known role for DNA methylation in promoting memory-like enhanced and sustained T cell function;(Akbari et al., 2021, Ladle et al., 2016, Wang et al., 2021) similar results were observed in a pre-clinical mouse model.(Jain et al., 2021) Deletion of DNMT3A in CAR T cells also promotes cell survival, limits exhaustion, and enhances therapeutic efficacy.(Prinzing et al., 2021) One study suggested that CHIP mutations may influence CAR T therapy as patients with CHIP had higher rate of complete response to CAR T therapy, but with increased rates of cytokine release syndrome.(Miller et al., 2021b) Aside from CHIP mutations in CAR T cells themselves, tumor-infiltrating leukocytes harboring CHIP mutations have also been observed in solid tumor samples (Kleppe et al., 2015). Thus, CHIP could impact the efficacy of immunotherapy via non-cell autonomous mechanisms that have not yet been explored.

Smoking

Cigarette smoking induces cytotoxic DNA damage and causes substantial tissue damage and inflammation. Smoking damages the HSC niche and subsequently impairs HSC homing, causes proliferative exhaustion of HSCs, and induces extramedullary hematopoiesis.(Morales-Mantilla et al., 2020, Siggins et al., 2014, Tura-Ceide et al., 2017) Increased risk of CHIP has been linked with a history of smoking,(Bolton et al., 2020, Genovese et al., 2014, Coombs et al., 2017, Kar et al., 2022) as well as smoking-associated pulmonary diseases such as COPD (Buscarlet et al., 2017, Zink et al., 2017) and lung cancer.(Zajkowicz et al., 2015, Coombs et al., 2017, Kessler et al., 2022) Among common CHIP mutations, mutations in ASXL1, have the most significant association with smoking status.(Bolton et al., 2020, Coombs et al., 2017, Kar et al., 2022, Kessler et al., 2022, Dawoud et al., 2020) Mutant ASXL1 regulates DNA transcription by enhancing the ubiquitinase BAP1, resulting in decreased histone ubiquitination and increased AKT/mTor signaling. This results in increased cell cycle progression even in the setting of DNA damage.(Fujino and Kitamura, 2020) Likely, smoking not only contributes to increased mutational burden of ASXL1 mutant clones but also provides a selective advantage for this clone via further induction of DNA and tissue damage and inflammation; further mechanistic studies of the relationship between smoking and ASXL1 mutations are still pending.

Inflammation is a common source of stress that induces the regulation and repair of physiological insults. Inflammation can be triggered by many sources, including old age, infections, cancer, radiation, and underlying pathogenic conditions like atherosclerosis. (Bick et al., 2020b, Gartung et al., 2019, Hansson et al., 2006, Matatall et al., 2016, Pietras, 2017) Inflammation is mediated by various signals, such as pathogen- and damage-associated molecular patterns (PAMPs and DAMPs), as well as pro-inflammatory cytokines including tumor necrosis factor alpha (TNFα), interferon alpha (IFNα), interferon gamma (IFNγ), IL-1β, and IL-6. As reviewed elsewhere, these signals influence HSCs via both direct and indirect mechanisms that impact quiescence, promote differentiation at the expense of self-renewal, and affect cell survival.(King and Goodell, 2011, Baldridge et al., 2010, Essers et al., 2009, Caiado et al., 2021, Pietras, 2017, Hormaechea-Agulla et al., 2021) Long-term consequences of chronic inflammation include impaired HSC self-renewal and subsequent HSC terminal exhaustion.(Matatall et al., 2016)

Several studies in both humans and animal models demonstrate that selected CHIP clones can resist inflammation-induced proliferative stress and gain a selective advantage in inflammatory conditions. Studies in patients with HIV(Bick et al., 2020b, Dharan et al., 2021) and aplastic anemia(Ogawa, 2016, Zhang et al., 2019, Yoshizato et al., 2015) indicate that inflammation can lead to the selection of specific CHIP clones, and can predispose to progression to AML and myelodysplastic syndrome (MDS).(Kristinsson et al., 2011)

Resistance to Inflammation-Mediated Depletion

Loss-of-function TET2 clones have been extensively studied in murine models and these TET2 mutants selectively expand, particularly in the myeloid compartment, in response to pro-inflammatory stimuli TNFα and lipopolysaccharide.(Abegunde et al., 2018, Cai et al., 2018) In a key study by Meisel et al., the presence of Tet2 null hematopoietic cells disrupted intestinal barrier integrity and increased systemic microbial signaling, particularly from Lactobacillus, which was critical in promoting further clonal expansion.(Meisel et al., 2018) Conversely, broad-spectrum antibiotic therapy slowed the expansion of Tet2-deficient clones.(Zeng et al., 2019) Mechanistically, the long noncoding RNA (IncRNA) Morrbid and the TLR4/NF-κB/IL-6/Stat3 pathway may be crucial for Tet2 clonal expansion and persistence.(Cai et al., 2018) Pharmacological inhibition of Shp2 with inhibitor SHP099 or inhibition of Stat3 with E3330 led to a loss of Morrbid hyperactivity in Tet2-deficient hematopoietic stem and progenitor cells (HSPCs), suppressing selective clonal expansion.(Cai et al., 2018) This study highlights the role of IncRNAs in modifying cell-death to inflammation by making them more resistant to inflammation-induced apoptosis.

While DNMT3A mutations are not as extensively studied as TET2, our recent findings shed light on mechanisms by which inflammation can positively select for Dnmt3a KO clones. We found that Dnmt3a KO murine clones outcompete wildtype murine cells in response to chronic IFNγ stimulation caused by Mycobacterium avium infection.(Hormaechea-Agulla et al., 2021) Much like TET2 clones, chronically IFNγ-stimulated Dnmt3a null HSCs are resistant to apoptosis, as caspase 3/7 activity is significantly decreased compared to IFNγ-stimulated wildtype HSCs. Most strikingly, Dnmt3A clones showed dampened differentiation in response to IFNγ compared to wildtype cells. Since we previously showed that excessive IFNγ-stimulated myeloid differentiation is a major mechanism of HSC depletion during chronic infection,(Matatall et al., 2016) reduced IFNγ-dependent differentiation provides a clear selective advantage to Dnmt3a KO clones. Indeed, Dnmt3a KO clones do not upregulate pro-differentiation genes, nor do they display significant myeloid differentiation upon IFNγ stimulation in vitro. Consistent with this phenotype, the promoter regions of pro-differentiation genes were hypermethylated in Dnmt3a KO HSCs from infected mice compared to WT.(Hormaechea-Agulla et al., 2021) Further, a recent study showed that IFNγ-dependent DNMT3A^mut clonal expansion was attributable to DNA hypomethylation and altered expression of Txnip and p21, allowing preservation of HSC quiescence.(Zhang et al., 2022) These findings demonstrate that epigenetic changes underlie transcriptional activation of inflammation-driven HSC differentiation and provide a clear mechanism by which Dnmt3a KO clones are selected to expand in inflammatory conditions.(Hormaechea-Agulla et al., 2021, Zhang et al., 2022)

While these studies provide a potential mechanism by which CHIP mutants, such as DNMT3A, expand in the setting of inflammatory stress, some parallel studies of humans and human cells have not mirrored exactly these findings. For instance, the lineage output of naive Dnmt3a KO HSCs in our murine model differed from what has been observed in human DNMT3A R882 mutated HSPCs.(Nam et al., 2022) Single-cell multi-omics of DNMT3A R882 CD34+ HSC populations in multiple myeloma patients revealed a myeloid biased differentiation and megakaryocytic expansion within humans, suggesting that broad transcriptional consequences of DNMT3A mutations affect differentiation even in the absence of a strong exogenous source of inflammation.(Nam et al., 2022) Altered methylation of human DNMT3A R882 mutant clones showed aberrant methylation status, but the differentially methylated regions included MYC as opposed to the ATF transcription factors that we highlighted. Thus, while murine models can shed light on the mechanisms by which DNMT3A mutant clones are positively selected during inflammation, mechanistic studies in humans must be done to confirm relevance.

Depletion of wild-type competitors

Other CHIP clones encode transcriptional alteration(s) that promote positive selection and suppression of competing wild type (WT) clones.(Higa et al., 2021) Our recent study showed that Cebpa KO MPP3 clones expand in response to chronic IL-1ß signaling. Transcriptionally, Cebpa deficiency prevents IL-ß-driven repression of self-renewal genes like Foxo3 and Mycn, which are directly regulated by Cebpa, thereby giving mutant clones a self-renewal advantage over WT cells. Interestingly, Cebpa KO clones can also suppress WT competitors by promoting excessive myeloid differentiation in a competitive transplant model. Indeed, RNA-seq analysis demonstrated that WT MPP3s exposed to chronic IL-1ß and in contact with Cebpa KO clones showed a significant increase in myeloid differentiation genes compared to WT MPP3s stimulated by chronic IL-1ß alone.(Higa et al., 2021)

Avagyan et al 2021 expanded upon the ability of mutant clones to promote differentiation of their competitors. Utilizing a novel technique called Tissue editing With Inducible Stem cell Tagging via Recombination (TWISTR) in a zebrafish model, Avagyan et al created a mosaic zebrafish that expressed 12 mutated CH genes during development. Interestingly, clones bearing frameshift mutations in asxl1, the zebrafish homolog of human ASXL1, demonstrated clonal dominance. Single-cell RNA sequencing of marrow cells showed asxl1 mutant macrophage and neutrophil clones were highly inflammatory, with upregulation of inflammatory cytokines il1β and tnfβ. More strikingly, the asxl1 mutant progenitors showed an inflammation-resistant molecular profile, upregulating genes that modulate the effects of inflammatory stress such as socs3a and nr4a1. These findings suggest that CHIP clones contribute to the generation of inflammatory cytokines known to deplete WT HSCs. At the same time, the clonal mutant progenitor cell types resist the deleterious effects of these inflammatory signals and expand.(Avagyan et al., 2021) It is interesting to note that some CHIP mutations (e.g. in MYD88, STAT3) play a key role in transmitting inflammatory signaling, suggesting a direct route for inflammation resistance. These findings highlight how mutant HSCs can promote their own selection by preserving self-renewal while WT competitors are outcompeted or exhausted through excessive myeloid differentiation.

CHIP as a cause and consequence of inflammation

While CHIP mutants have survival and selective advantages in response to inflammation, CHIP itself also drives inflammation, setting up a feed-forward loop, as noted in the above example of asxl1-mutant zebrafish and reported in association with other mutations. For instance, PPM1D and TET2 mutant clones have been shown to produce immune cell progeny that generate significantly higher than normal levels of the pro-inflammatory cytokine IL-6 upon exposure to chemotherapy or lipopolysaccharide, respectively.(Cai et al., 2018, Jaiswal et al., 2017, Yura et al., 2021) TET2-mutant clones contribute to atherosclerosis and heart disease via increased production of IL-6 by downstream macrophages, and DNMT3A has recently been suggested to contribute to inflammation in the heart by the same mechanism. (Abplanalp et al., 2021, Bick et al., 2020a) As previously noted, TET2 mutations may also contribute to systemic inflammation by causing intestinal barrier compromise. As genetic mosaicism is now recognized to occur widely across tissues,(Hsieh et al., 2020) additional proinflammatory consequences of CHIP mutations are likely to arise.

Human studies support the notion that CHIP can contribute to inflammation-mediated diseases such as severe COVID-19, sepsis, and cardiovascular disease.(Bolton et al., 2021, Dawoud et al., 2020, Jaiswal and Ebert, 2019, Kessler et al., 2022, Zekavat et al., 2021b) While increased IL-6 has been linked to heart disease in animal models and blockade of IL-6 signaling can reduce cardiovascular risk,(Kessler et al., 2022) no empirical evidence has been published suggesting that IL-6 itself is capable of promoting CHIP. Using the presence of the IL6R p.Asp358Ala coding mutation as a proxy to assess the potential for IL-6R blockade in individuals with CHIP, Bick et al., 2020 found that impaired IL-6R signaling reduced cardiovascular disease risk but did not impact CHIP status, particularly DNMT3A and TET2 CHIP. Furthermore, transcriptomic studies have not shown upregulation of canonical inflammatory pathways in CHIP, and the role of IL-6 signaling in CHIP-associated CVD risk remains an open question.(Kessler et al., 2022, Weinstock et al., 2021) Thus, while mechanistic studies in murine models highlight ways that inflammation drives CHIP, the role of inflammation as a driver of CHIP in humans requires further experimental validation. Such a role may be even more complex to identify in humans due to the impact of cell-intrinsic genetic modifiers such as TCL1A on the selective expansion of CHIP clones. (Weinstock et al., 2021)

The extent to which inflammation suppresses competitor cells or changes the niche to allow for favorable expansion of CHIP clones has yet to be fully explored.(Rodriguez-Meira et al., 2022) While we have shown that CHIP clones resist the differentiation and exhaustion effects of inflammation, future studies could highlight more active processes by which CHIP clones suppress their competitors.(Rodriguez-Meira et al., 2022)

Autoimmunity

Autoimmune disease mediated in part by dysregulated cytokine production, including IL-1, IFNγ, and TNFα,(Italiani et al., 2018, Muzes et al., 2012, Lubberts and van den Berg, 2003) generates a chronic inflammatory environment that could facilitate the expansion of selected CHIP clones. Autoimmune diseases like inflammatory bowel disease, systemic lupus erythematosus, and rheumatoid arthritis are well known to carry increased risk of myeloid malignancies.(Bekele and Patnaik, 2020, Boddu and Zeidan, 2019, Ramadan et al., 2012) The risk of hematological malignancy in these diseases has often been attributed to the use of therapies such as cyclophosphamide, methotrexate, and TNFα blockade.(Ramadan et al., 2012) However, increased risk of leukemia has been observed in patients with arthritis with no prior history of treatment, suggesting inflammation itself may have a role in hematological malignancy risk.(Boddu and Zeidan, 2019) One small study found an increased prevalence of CHIP mutations in arthritis patients, with higher VAFs correlated with more severe disease, although this was not replicated in a separate study.(Tariq et al., 2020, Savola et al., 2018) Another small cohort study of inflammatory bowel disease patients found increased VAF for DNMT3A and PPM1D compared to prior cohort studies, which was independent of treatment history. Additionally, serum levels of IFNγ were elevated in patients with DNMT3A^mut CH,(Zhang et al., 2019) further implicating this inflammatory cytokine in the selection of DNMT3A^mut clones. Indeed, altered epigenetic profiles are known to facilitate immune dysfunction in autoimmune diseases.(Mazzone et al., 2019) Altogether, mutations in epigenetic modifiers could provide a selective advantage for CHIP expansion and subsequent autoimmune disease progression. It is important to note that most of these studies on autoimmune diseases are in small cohorts that are underpowered. Further mechanistic studies are warranted to elucidate the role of autoimmunity and its treatment in CHIP expansion.

Hematopoietic Stem Cell Transplants

HSCT is a last-resort treatment for hematological diseases like AML, MDS, aplastic anemia, and SCD. However, the stress of HSCT promotes loss of HSC proliferative and regenerative capabilities(Flach et al., 2014, Harrison and Astle, 1982) through telomere shortening(Colla et al., 2015, Notaro et al., 1997) and epigenetic changes,(Soraas et al., 2019) decreasing overall HSC fitness. Furthermore, preconditioning for bone marrow transplant via irradiation and cytotoxic regimens can damage the niche(Pinho and Frenette, 2019) and generate an inflammatory environment, further impairing HSPC fitness. Several studies have demonstrated that some CHIP clones have a selective repopulation advantage during transplantation stress.

Both murine and human studies have shown that TET2-deficient HSCs have increased self-renewal and myeloid-biased repopulation expansion after transplantation.(Kunimoto and Nakajima, 2020, Moran-Crusio et al., 2011, Ortmann et al., 2019, Ostrander et al., 2020, Shide et al., 2012) A study by Abegunde et al. showed that TET2 mutant clones avoid HSC suppression by inhibiting TNFα signaling.(Abegunde et al., 2018) However, contradictory studies suggest that TET2^null HSCs unexpectantly exhaust at the same rate as control HSCs in serial transplantation.(Ostrander et al., 2020) Furthermore, Yamashita et. al. found that TNFα plays a key role in facilitating HSC survival and myeloid differentiation;(Yamashita and Passegue, 2019) therefore, if TET2 loss of function inhibits TNFα signaling one might expect impaired HSC survival. Overall, the precise mechanism of TET2-mutant clonal expansion after transplantation remains poorly defined and may involve pathways other than TNFα.

Murine studies have shown that Dnmt3a mutant clones have virtually limitless self-renewal and competitive repopulating capacities,(Challen et al., 2011, Kunimoto and Nakajima, 2020) allowing them to outcompete their normal counterparts over time, at least in the context of stem cell transplantation after irradiation preconditioning. Indeed, DNMT3A loss-of-function HSCs can regenerate over at least 12 transplant generations despite the stress of serial transplantations.(Jeong et al., 2018) Allsopp et. al. suggested that DNMT3A^null embryonic stem cells have elongated telomeres that can extend the replicative lifespan of serially transplanted HSCs.(Allsopp et al., 2003, Gonzalo et al., 2006) Resistance to inflammation-mediated decay may also contribute to persistence of Dnmt3a-mutant clones after transplantation, a topic of ongoing exploration.

In contrast to TET2 and DNMT3A, ASXL1, SRSF2, SF3B1, U2AF1, and PPM1D are unable to outcompete wildtype HSPCs in murine transplant settings.(Fujino et al., 2021, Yoshida et al., 2011, Hsu et al., 2018, Kon et al., 2018, Mupo et al., 2017) PPM1D mutant clones lack proliferative advantage and have overall decreased fitness after HSC transplantation, as shown in a series of patients after autologous transplant.(Hsu et al., 2018, Wong et al., 2018) These studies suggest that stabilization of the DDR and resistance to apoptosis conferred by PPM1D mutations confer little to no benefit in the context of the proliferative stress of HSCT.(Hsu et al., 2018, Kahn et al., 2018, Wong et al., 2018)

While gain-of-function mutations in PPM1D were not selected following HSCT, TP53 mutant HSC clones have a competitive expansion advantage under transplant stress.(Chen and Liu, 2019, Chen et al., 2019) Chen et al. demonstrated that a gain-of-function mutation allows TP53 to have enhanced interaction with EZH2, reinforcing H3K27 trimethylation and thereby enhancing the mutant clone’s self-renewal and differentiation capacity.(Chen and Liu, 2019, Chen et al., 2019) Interestingly, TP53 HSC mutant expansion also occurs in a non-irradiated transplant model in absence and, to a greater extent, in the presence of doxorubicin chemotherapy.(Sano et al., 2021) Thus, both mechanisms of altered epigenetics and DDR pathway dysfunction may facilitate selective advantage of TP53 mutant clones due to both transplantation and chemotherapy stress.

CHIP has been observed in patients following autologous (Chitre et al., 2018, Gibson et al., 2017b) and allogeneic stem cell transplants (Frick et al., 2019) and may affect stem cell repopulation and engraftment. Accumulating evidence shows that HSCs harboring CHIP mutations can be found in the donor and can be transplanted to the recipient (Gibson et al., 2017a, Nawas et al., 2021). Recipients who received donor grafts with CHIP mutations experience a more biased myeloid lineage expansion, increasing the risk of the development of adverse hematological outcomes, including therapy-related myeloid neoplasms and cytopenias.(Chitre et al., 2018, Gibson et al., 2017a, McNerney and Le Beau, 2018) Furthermore, Pierola, et al. showed in an in vivo study that CHIP mutant HSC clones lacking DNA damage response systems can survive chemotherapy-induced stress and DNA damage while acquiring secondary mutations, ultimately leading to secondary malignancies. (Alfonso Pierola et al., 2016) Altogether, these discoveries are leading to increased scrutiny of the mutational landscape in HSCT donors.

Aging

Aging is the stressor that is most strongly associated with CH.(Bolton et al., 2020, Genovese et al., 2014, Jaiswal et al., 2014, Xie et al., 2014) Sequencing studies in the elderly reveal that clonal hematopoiesis is most prevalent in those aged >70. While many CH clones can emerge, the most prevalent age-associated clones are DNMT3A, TET2, ASXL1, SF3B1, and SRSF2.(Buscarlet et al., 2017, Genovese et al., 2014, McKerrell et al., 2015, Jaiswal et al., 2014, Xie et al., 2014)

While aging can lead to the expansion of various CHIP clones, it is noteworthy that the expansion of these clones is not directly correlated with CHIP-associated clinical outcomes. A study by Zeventer et al., 2021 investigated CHIP in individuals over 80 years old and found no association between cardiovascular morbidity and the presence of CHIP. There were, however, associations between CHIP and deaths due to hematological malignancies (hazard ratio 1.48, CI 95%).(van Zeventer et al., 2021) Of note, the most prevalent clones in this study were DNMT3A and TET2 mutants, both of which are associated with cardiovascular diseases.(Jaiswal et al., 2017) These opposing findings indicate that aging can lead to the expansion of clones but that the major impacts on health occur before a threshold age, suggesting that the age CHIP onset may be the most relevant factor.

Mechanistically, aging can be considered a multi-factorial stressor that leads to various effects on the hematopoietic compartment, as explored in several reviews.(Jaiswal and Ebert, 2019, Lee et al., 2019, Libby et al., 2019) The mechanisms underlying age-associated clonal expansion likely overlap with inflammatory stimuli discussed above. Indeed, there is a baseline increase in production of pro-inflammatory cytokines such IFNγ, TNFα, and IL-6(Chung et al., 2006, Li et al., 2011, Sanada et al., 2018) over time, frequently called “inflammaging,” which could promote the expansion of DNMT3A(Hormaechea-Agulla et al., 2021) and TET2(Abegunde et al., 2018, Meisel et al., 2018) CHIP clones. In addition, aged HSCs have increased histone and DNA methylation, resulting in activation of self-renewal signatures and silencing of differentiation programs.(Sun et al., 2014, Nachun et al., 2021, Beerman, 2017) As a result, aged HSCs display defective homing to the bone marrow niche, impaired lymphocytic differentiation, and increased myeloid and platelet lineage priming.(Franceschi and Campisi, 2014, Kovtonyuk et al., 2016, Lee et al., 2019) This overproduction of myeloid cells may contribute to another feed forward loop of increased inflammation and CHIP during aging.(Franceschi et al., 2000)

Interestingly, Kar et al., 2022 recently reported that longer leukocyte telomere length was associated with increased DNMT3A- and TET2-mutant CHIP.(Kar et al., 2022) Long telomeres may be yet another mechanism by which these clones persist despite inflammation and replicative stress, but this proposed mechanism has not yet been formally demonstrated.

A study by Fujino et al., 2021 showed that C-terminally truncated ASXL1 clones expanded in aged mice. (Fujino et al., 2021) Mechanistically, truncated Asxl1 interacts with BAP1 to deubiquitinate Akt and activate mTORC, leading to the survival and expansion of these aged Asxl1 HSCs. However, Asxl1 clones exhibited impaired engraftment, myeloid-biased hematopoiesis, and significantly decreased homing ability to the bone marrow niche compared to wildtype transplanted HSCs, consistent with the observation that ASXL1 clones do not survive the stresses of HSCT.(Fujino et al., 2021)

Other investigators have used modeling to argue that a constant fitness advantage conferred by the CHIP mutation is sufficient to model the CHIP seen in humans.(Fabre et al., 2021, Watson et al., 2020) In fact, Fabre et al conclude that “many clones grew more rapidly early in life compared with the rate in old age”.(Fabre et al., 2021) In contrast, Monte Carlo modeling has been used to show that cancer-causing mutations have greater positive selection late in life, thus explaining the age-dependence of leukemias and other cancers.(Rozhok and DeGregori, 2019) Additional studies, such as using mouse and non-human primate models, will be required to determine whether CHIP occurs simply as a matter of time or if extrinsic pressures are required.

Metabolic Stress

Metabolism plays a crucial role in maintaining and regulating HSC homeostasis,(Ito et al., 2019)) and there has been increasing interest in uncovering how metabolic factors impact CHIP. In particular, TET2 mutant clones have been known to interact with endocrine and metabolic axes, although the reports of these associations have been mixed. In a type II diabetes mouse model, a TET2 mutant clone was shown to facilitate insulin resistance in mice.(Fuster et al., 2020) In addition, elevated glucose levels have been found to destabilize TET2 in human peripheral blood samples,(Wu et al., 2018) and increased glucose levels drove leukemia formation in mice harboring TET2-mutations.(Cai et al., 2021) Two recent large cohort studies found a higher prevalence of CHIP in people with an unhealthy diet(Bhattacharya et al., 2021) and in postmenopausal women with a high BMI and poor diet quality.(Haring et al., 2021) A longitudinal study of 20 patients showed that clonal expansion was associated with lower high-density-lipoprotein (HDL) cholesterol and insulin resistance but not with BMI, hypertension, hyperglycemia or total cholesterol.(van Deuren et al., 2021) However, UK BioBank data revealed hypertension, low total cholesterol, and low LDL to be associated with CHIP but not obesity or diabetes.(Kar et al., 2022) A separate analysis of UK BioBank data showed that elevated BMI was negatively associated with DNMT3A mutation, but positively associated with TET2 and ASXL1 mutant clones.(Kessler et al., 2022) One confounder that may explain these mixed results is that vitamin C has been shown to regulate HSC homeostasis by activating TET2.(Lee et al., 2020) Lower plasma vitamin C concentrations was associated with a higher frequency of TET2 mutations in an elderly cohort.(Chen et al., 2021) For DNMT3A, murine models showed that a fatty-bone marrow environment can provide a selective advantage for DNMT3A-^mut cells which may be mediated by IL-6.(Zioni et al., 2021) Altogether, early studies suggest an important role for metabolic and dietary stresses in the pathogenesis of CHIP.

Other Stresses That Expand CHIP Clones

While we focused in this review on specific DNA-damaging and inflammatory environmental drivers of clonal expansion, CHIP has also been linked with various other potential drivers including psychiatric diseases,(Zink et al., 2017) gout,(Kessler et al., 2022) and Down syndrome.(Liggett et al., 2021, Tong et al.). In individuals with Down syndrome, clonal hematopoiesis is observed much earlier in life with the most common mutations being found in TET2; although notably this CHIP did not reach levels observed in the elderly.(Liggett et al., 2021) Trisomy 21 has been shown to induce an enhanced IFN signature in the hematopoietic compartment, which may facilitate early clonal expansion,(Sullivan et al., 2016) and CHIP in individuals with Down Syndrome is associated with dysregulated immune patterns in peripheral leukocytes.(Liggett et al., 2021) Selective clonal expansion in other diseases like SCD is less clear.(Pincez et al., 2021, Liggett et al., 2022) A recent report suggests that space flight may accelerate clonal expansion,(Mencia-Trinchant et al., 2021) and many more associations and drivers of CHIP are constantly being discovered.

Concluding Remarks

The natural acquisition of mutations throughout life generates a range of genetically distinct HSPCs. Despite the ubiquity of CHIP-associated mutations, most individuals never develop CHIP, suggesting that environmental pressures are critical determinants of clonal expansion. (King et al., 2020) Here we reviewed the contexts in which some genetically variant clones expand over time. We highlight specific examples of imposed external stress such as platinum-based chemotherapies that drive the expansion of clones containing PPM1D mutations, which in turn increase the risk of secondary malignancy. As the field increasingly identifies environmental factors that may provide a selective advantage for certain clones, the clinical implications of these associations and whether they can be modulated to improve human health continue to be explored.

Among genes known to be associated with CHIP, DDR factors such as P53 and PPM1D have been associated with chemotherapies and radiation. Mutations affecting epigenetic modifiers — DNMT3A, TET2, ASXL1 — and JAK2 have been associated with bacterial translocation, infections, and inflammatory conditions, and one would expect that MYD88 and STAT3 mutant clones are similarly selected in the setting of inflammation. Relatively little is known about the environmental conditions that drive expansion of clones containing mutations in splicing factors such as SF3B1, SRSF2, and U2AF1, which are among the top 10 genes mutated in CHIP. These mutations have been reported to drive inflammatory responses, but further work will be required to determine if they follow the pattern of inflammation-driven selection that has been reported for TET2, DNMT3A, ASXL1, and JAK2.

Whereas some variant clones may contribute to heart disease or stroke, others may be neutral in their impact on human health or may be beneficial by enabling long-term hematopoiesis in the elderly. Indeed, despite our tendency towards binary thinking, mutant clones may live in the liminal space between pathogenic and beneficial. For instance, whereas TET2 clones drive increased IL-6-mediated inflammation, they may also enable persistent hematopoiesis in the challenging conditions of advanced age or serve to suppress competing clones with more serious adverse health effects, essentially constituting non-malignant ‘fitness peaks’. While mutant VAF >10% is strongly associated with increased heme malignancy risk, what other factor(s) govern progression to pre-malignant states like clonal cytopenia of undetermined significance (CCUS) and/or leukemia in this setting remain to be clearly identified. For some mutations, downstream consequences of mutations on differentiated cell function may have an outsized effect on health. Further, most mechanistic studies to elucidate environmental drivers of CHIP have been done in mice. While informative, these models have limitations, the most obvious of which is the short lifespan of the mouse relative to humans and the clean (if not sterile) and non-challenging housing conditions. Paired epidemiologic association studies have been a tremendous strength in this respect, but utilization of clinical trials and non-human primate studies will also be important.

While a range of studies highlights how the environment (e.g., obesity, infection, prior exposure to chemotherapy) significantly alters which clones are selected to survive and expand, the extent to which changes in environment can shape clonal selection or modify disease risk remains to be seen. Lessons from the CANTOS trial examining IL-1ß inhibition to reduce cancer risk are a case in point: while greatly reducing lung cancer risk and potentially reducing major adverse cardiovascular events in patients with TET2-mutant CHIP, canakimumab leaves patients vulnerable to potentially deadly infections. (Ridker et al., 2017, Svensson et al., 2022) Outcomes from such trials indicate that endpoints for intervention must be carefully defined. Furthermore, basal inflammatory signaling, such as is mediated by the microbiome,(Josefsdottir et al., 2017) is essential for normal hematopoiesis and must be taken into account when considering anti-inflammatory approaches. Perhaps it is more productive to address the aberrant immunometabolic features of immune cells derived from mutant clones than to reduce the size of the clone itself. Similarly, one can consider more nuanced methods (beyond brute inhibition of inflammation) to modulate the bone marrow microenvironment to reduce the selective advantage of certain CHIP clones. Finally, while an intervention may prevent the expansion of a specific CHIP clone, it also may enhance the emergence of an alternate clone with unknown impact on health.

Evidence that PPM1D mutant clones expand only in the context of certain chemotherapies raises the potential of using personalized medicine approaches in the selection of therapy. One can imagine an array of decision-making algorithms to guide therapeutic approaches to cancer or other conditions while taking a patient’s somatic mutational background into account. Innumerable studies, including an assignment of the relative risk of various mutations and mutant genes, will be needed to make such a personalized approach truly evidence based.

The relative accessibility of blood enabled the discovery of CHIP and its health implications. Recent studies show that clonal expansion also occurs in other tissues and may be a precursor to solid tumors.(Kakiuchi and Ogawa, 2021, Wijewardhane et al., 2021) Non-oncologic health consequences of clonal expansion in other tissues have yet to be defined, and knowledge in this area will undoubtedly expand in the coming years. These future studies will shape our understanding of how systemic therapies or lifestyle modifications impact clonal evolution in non-hematologic tissues.

King and colleagues analyze how environmental contexts associated with clonal hematopoiesis of indeterminate potential (CHIP) alter tissue microenvironments to facilitate the selection and expansion of specific CHIP mutant clones. They also consider major remaining gaps in knowledge that will determine future application of this field in transplant and preventive medicine.