Clonality in immune aplastic anemia: mechanisms of immune escape or malignant transformation

Abstract

Aplastic anemia (AA) is the prototypic bone marrow failure syndrome and can be classified as either acquired or inherited. Inherited forms are due to the effects of germline mutations, while acquired AA is suspected to result from cytotoxic T-cell mediated immune attack on hematopoietic stem and progenitor cells. Once thought to be a purely “benign” condition, clonality in the form of chromosomal abnormalities and single nucleotide variants is now well recognized in AA. Mechanisms underpinning this clonality likely relate to selection of clones that allow immune evasion or increased cell survival the marrow environment under immune attack. Wide spread use and availability of next generation and other genetic sequencing techniques has enabled us to better understand the genomic landscape of aplastic anemia. This review focuses on the current concepts associated with clonality, in particular somatic mutations and their impact on diagnosis and clinical outcomes in immune aplastic anemia.

Introduction:

Since its first presentation in the late 19^th to early 20^th century, Aplastic anemia (AA) has served as the prototypic bone marrow failure syndrome(1). It manifests as cytopenias with a hypocellular bone marrow and consequent clinical sequelae. Aplastic anemia can be classified as either immune/acquired or inherited. Inherited bone marrow failure syndromes (IBMFS) result from germline (GL) mutations and are often identified at birth or in early childhood, but also less frequently later in life(2, 3). An increased risk of malignant transformation is reported with IBMFS, requiring life-long monitoring. Immune aplastic anemia (AA) on the other hand is diagnosed using the Camitta criteria after exclusion of IBMFS or direct toxicity to the marrow. Immune AA has been the center of much speculation, and although exact mechanisms and triggers remains elusive, experts agree that immune dysregulation resulting in an autoimmune T-cell regulated destruction of hematopoietic and progenitor stem cells (HSPCs) is the likely driver(4). This hypothesis is supported by the consistently observed responses to immunosuppressive agents such as anti-thymocyte globulin (ATG) and cyclosporin (CSA).

Traditionally, immune AA has a bimodal presentation with the first peak between the ages of 15–25 and the second > 60 years. It is a rare disease with incidence ranging from 2–14 per million per year. The severe and very severe (defined as ANC <200 cells/μL) forms of the disease are typically fatal if left untreated. With a progressive and methodical understanding of the pathophysiology of aplastic anemia over the last 30–40 years, the management of AA has evolved from supportive to therapeutic. Most patients can now be successfully treated with a combination of immunosuppressive therapy (IST) and Eltrombopag (EPAG) or allogeneic bone marrow transplants (BMT). For younger patients BMT is considered in the first line setting when there is a fully matched sibling or matched unrelated donor (MUD). Increasingly haplo-identical transplants are being used in AA in the relapsed / refractory setting if no matched sibling or MUD donors are available; studies are ongoing assessing their role in treatment naïve patients(5–7). IST is an excellent option for older patients or younger patients without a fully matched donor, with response rates approaching 80% in the treatment naïve setting. However, in patients treated with IST there remains a well-established long-term risk of clonal evolution to myeloid neoplasia, as well as a strong clinical association with the development of paroxysmal nocturnal hemoglobinuria (PNH). It has become apparent that many AA patients exhibit some form of clonality – be it structural aberrations or somatic gene mutations. Likely, this relates to the specific marrow environmental pressure present in immune AA favoring clones that can either evade the immune system or are more resistant to cell destruction. In this review we will explore the role of clonality and somatic mutations in AA.

Clonality in aplastic anemia

Mutations acquired from the embryonic stage and through later life are recognized as somatic; they are not inherited or passed on to progeny. Somatic alterations are increasingly recognized as contributing to AA pathogenesis. However, higher throughput and sensitivity, and a wider application of clinical sequencing for the detection of somatic mutations has also resulted in new challenges in how to integrate these findings into clinical practice. For example, the presence of clonal hematopoiesis (CH) in AA patients at low levels, and the detection of variants of uncertain significance of germline and somatic origin present diagnostic challenges(8). Age related clonal hematopoiesis (ARCH) is a common phenomenon observed in the healthy aging population where HSPCs give rise to genetically distinct subpopulation(s) of blood cells. One important determination is whether the presence of CH in AA is a pathogenic finding or whether it can be a benign or neutral consequence of the underlying disease. Clearly, AA patients show an enriched mutational frequency compared to healthy age matched individuals. There are two suggested hypotheses for this observation. First, the pre-existing or new HSCs with mutations gain a selective survival and proliferative advantage over the normal HSCs in the immune mediated cytotoxic environment as the marrow recovers--an immune escape mechanism. Second, the mutation(s) are, in fact, the original antigenic target of the immune mediated attack but normal HSCs are obliterated by the virtue of being in the same vicinity as has been postulated in the case of PIGA. As the marrow recovers, somatic mutations occur and persist in the recovering HSC compartment.

Clinically, the course of AA may be complicated by the development of PNH (approx. 30–60% of cases)(9–11), or less frequently, clonal evolution to myelodysplastic syndrome (MDS) or acute myeloid leukemia (AML) (approx.10–15%)(12, 13). Clonality may also be present due to loss of HLA alleles(13, 14), the presence of somatic mutations in myeloid cancer genes or PIGA, or the expansion of clonal T cell populations. Classically, malignant clonal evolution has been defined by the acquisition of an abnormal karyotype or morphological MDS or AML, however, it is being increasingly recognized that particular somatic mutation drivers may play an important role in both the diagnosis and prognosis of this secondary MDS. Somatic mutations in clonal hematopoiesis of indeterminate potential (CHIP) or myeloid cancer genes have been detected in 30% of AA patients at baseline and up to 60% after IST(15). The pattern of somatic mutations is frequently unpredictable and not all mutations are associated with disease development nor considered to be driver mutations. It is therefore imperative to understand the impact of acquired mutations as individual, as well as in combination with genetic aberrancies. It also pertinent to differentiate between the presence of small PNH clones from clinically symptomatic PNH with hemolysis requiring therapeutic interventions(16).

Cytogenetic abnormalities in AA

Aplastic anemia originally was thought to be devoid of any clonal abnormalities. However, AA patients have long been known to develop either overt hematologic malignancy or isolated abnormal karyotypes subsequent to IST, and it is now recognized that a subset of patients may present with abnormal cytogenetics, most commonly 13q deletion (del 13q). While del 13q is an MDS defining cytogenetic abnormality, when present in a patient with a hypocellular marrow and no morphological dysplasia it is considered by most experts AA rather than MDS(17). Del 13q, along with trisomy 8, are also often seen in AA patients after treatment with IST and are thought to develop due to immune escape mechanisms. There is a competitive advantage afforded to the HSCs harboring trisomy 8 via the induction of T-cell suppression. This is in response to the increased WT-1 peptide antigen expression resulting from the suppression of non-trisomy 8 HSCs(18). However, the immune escape mechanism in del 13q is not very well understood. Both trisomy 8 and del 13q indicate a higher response rate to treatment with IST(9, 19–21).

Monosomy 7 or deletion 7q aberrations are responsible for most AA clonal evolution to MDS/AML (63%)(21, 22); they are enriched when compared to primary MDS where the frequency of monosomy 7 or deletion 7q is reported to be only 14%(22). Patients with monosomy 7 have a high risk of progression to leukemia, even if no morphological myeloid neoplasia is present at its first detection, and have a dismal overall survival in the absence of BMT(23). Patients who develop monosomy 7 do not exhibit high numbers of somatic mutations, and the predominance of chromosome 7 abnormalities in immune AA may relate to accelerated telomere attrition in the setting of a low HSC pool(24).

Other less common alterations include trisomy 6 and 15; in the absence of morphological dysplasia these likely represent low-risk clonal events(23)

Somatic mutations in AA, clonal hematopoiesis of indeterminate significance (CHIP) and myeloid neoplasms

CHIP is defined as the presence of a somatic mutation in a known driver gene at variant allele fraction (VAF) ≥2% and serves as a prototypic example of the presence of mutations (often at a low VAF of <10%) without clinically significant cytopenias(25). The three most frequent mutations generally reported are TET2, ASXL1 and DNMT3A. Other reported mutations include PPM1D, JAK2, SF3B1, SRSF2, TP53, GNAS and GNB1. While cytopenias are usually not seen in individuals carrying CHIP mutations, there have been other associations observed such as increased inflammation and cardiovascular disease(8, 26, 27). CHIP is associated with a 0.5–1% per year risk of progression to leukemia(8, 25). When these driver mutations are present with cytopenias, patients are deemed to have clonal cytopenia of undetermined significance (CCUS) and have a much higher known rate of progression to MDS(28).

CH mutations are commonly detected in AA patients. In the largest published sequencing study of 439 AA patients from the USA and Japan, somatic mutations occurred in up to 30% when assessed 6 months after IST. The most common somatic mutations were in BCOR/BCORL1 (9.3%), PIGA (7.5%), DNMT3A (8.4%), and ASXL1 (6.2%), together accounting for 77% of all the mutations in positive patients(13). Patients with BCOR/L1 and PIGA1 mutations were found to have a superior overall survival and better response to IST compared to both non-mutated patients or those with other mutations such as ASXL1 and DNMT3A. In the recently published RACE study, mutations were detected in approximately 60% of patients 6 months after IST. DNMT3A was the commonest mutation detected both at baseline and 6 months after IST, followed by PIGA and BCOR/L1. In contrast to CHIP seen in healthy individuals, patients with AA-related CH have higher rates of BCOR/L1 mutations and lower rates of TET2. Additionally, the landscape of somatic mutations differs between younger and older AA patients, with PIGA, BCOR and BCORL1 the predominant mutations up to the 20–30 year age group while past the age of 30 years non PIGA, BCOR and BCORL1 mutations were commoner(13).

The presence of somatic mutations in AA patients has been shown to predict for MDS. Initially this was demonstrated in a UK study of 150 patients where those with any somatic had a higher risk of MDS with ASXL1 shown to be enriched in the patients with clonal evolution(29). Recently, in a cohort of 407 patients from the National Institutes of Health, the presence of specific mutations were found to be predictive of evolution to MDS/AML when detected at 6 months after IST; these included RUNX1, splicing factor mutations, and ASXL1. In contrast, BCOR/L1 and DNMT3A were not predictive of clonal evolution(23). However, to date the presence of a somatic mutation in a patient who otherwise meets criteria for AA is not considered diagnostic of MDS rather than AA. Additionally, acquisition of a somatic mutation in an AA patient treated with IST is not currently considered evolution to MDS in the absence of an abnormal karyotype or morphological neoplasia on bone marrow examination.

While AA, MDS and AML share similarities in the pattern of somatic mutations, the clonal landscape does differ somewhat between them. ASXL1 and DNMT3A are frequently encountered in all three diseases. In AA without malignant evolution, PIGA, BCOR/BCORL1, and DNMT3A predominate while spliceosome mutations, TET2, JAK2, RUNX1 and TP53 are underrepresented. This is suggestive of discrete and selective evolutionary mechanisms in AA. In general, the presence of mutations in spliceosome factors (SF3B1, SRSF2, U2AF1), cohesion complex, PRC2 complex and especially TP53 and RUNX1 are more pathognomonic of a malignant clonal hematopoiesis(22, 30, 31). Spliceosome mutations associated with ring sideroblasts (SF3B1) and those associated with myeloproliferative features (SRSF2, ZRSR2) are typical driver mutations enriched in classic MDS but not in hypoplastic MDS or AA(22, 32). Comparisons of mutational spectra of AA to hypocellular MDS showed similarities in pattern, but BCOR/BCORL mutations were more frequently encountered in AA while TET2 mutations were less commonly observed(33, 34). Importantly, the somatic pattern in hypocellular MDS was distinct from that observed in secondary MDS evolved from AA, where mutations in RUNX1, SETBP1 and ASXL1 as well as chromosome 7 abnormalities were prevalent(22) Additionally, the mean allelic burden of mutations in acquired immune AA is substantially lower than that in MDS (9.3% vs 30.4%) and AML.

While somatic mutations alone are not currently diagnostic of MDS, it is possible that molecular data will in the future be incorporated into diagnostic criteria as a primary marker of MDS, and thus may exclude patients with specific mutations from an AA diagnosis even in the absence of morphological dysplasia. However, the clinical significance of a mutation at small VAF identified at diagnosis, using deep sequencing methods, is unclear and with improving sequencing techniques increasingly smaller clones will continue to be identified both in healthy individuals and in AA patients.

Myelodysplastic syndrome and aplastic anemia: cytogenetic and genomic landscape

Diagnosis of bone marrow failure syndromes can be challenging due to rarity, often young age at presentation, and overlapping clinical pictures of these disorders. Pancytopenia, macrocytosis and reticulocytopenia are the most common indicators of bone marrow failure and the presence of hyperplasia and overt dysplasia on bone marrow biopsies provide the pathological basis of distinguishing classical MDS from AA.

Chromosomal abnormalities are also key to differentiating AA from MDS. Cytogenetic abnormalities in AA are rare at approx. 10–15% and predominated by deletion 13q(29). The presence of chromosome 7 aberrations such as del 7q and monosomy 7, or presence of del 5q are pathognomonic of MDS and exclude a diagnosis of AA; when acquired after AA diagnosis chromosome 7 abnormalities represent clonal evolution to MDS. The shared marrow hypocellularity in both AA and in hypoplastic MDS points toward a shared immune mechanism responsible for deficient blood cell production. Hypocellular MDS is known to respond to IST though a high-risk karyotype such as monosomy 7 would preclude this treatment.

PNH/ PIGA in AA

Paroxysmal nocturnal hematuria is a bone marrow failure syndrome characterized by intravascular hemolysis and increased risk of thrombosis. First descriptions date back to the mid-19^th century with records of clinical manifestations documented by the English physician Sir William Gull. Post AA, 35–60% patients have a detectable PNH clone via flow cytometry and approximately 15% of these clones coincide with clinically symptomatic PNH(35). While smaller clones often do not require treatment, patients with clinical sequalae or clones >50%, which have been associated with thrombosis even in the absence of overt hemolysis, will require treatment with anticomplement therapy. Therapies include eculizumab and ravulizumab which target complement protein C5, and Pegcetacoplan a newer agent which binds to protein C3 and is now FDA approved for use as second line therapy.

Somatic mutations or loss of the functional PIGA gene (PIGA⁻) lead to loss of cell surface bound glycosylphosphatidylinositol (GPI) proteins CD55 and CD59(35, 36). This loss of expression results in the lost ability of the erythrocytes to inactivate surface complement and leads to red cell destruction and intravascular hemolysis. Anticomplement therapy, while very effective clinically, does not directly target the PIGA clone.

Interestingly, healthy controls are known to have preexisting PNH HSCs at a very low level. In immune AA it is suspected that the PIGA⁻ HSC have an intrinsic selection advantage and escape the T cell mediated immune destruction though the exact immunologic escape mechanism is unclear. The GPI-anchored surface proteins themselves may be antigenic stimuli and as PIGA mutations result in lack of this surface protein, this enables HSCs with PIGA mutation to escape the cytotoxic cell destruction(37). Switching of MHC II antigenic presentation to MHC I as a consequence of anchorless GPI protein has also been proposed as a mechanism (10). The clonal expansion in PNH is strongly linked to the presence of HLA-DR2(38). In addition, the enrichment of HLA-DR2 in both AA and PNH supports the hypothesis of an immune mediated pathophysiology as it is associated with good prognosis and response to IST(38).

BCOR/BCORL1 in aplastic anemia

BCOR tumor suppressor gene is involved in lymphoid development by potentiating BCL6 repression, maintaining pluripotency of embryonic stem cells, and regulating mesenchymal stem cell and hematopoiesis. The BCOR and BCORL1 genes are located on the X chromosome at band Xp11.4 and Xp26.1 respectively; they are reported to be mutated in approximately 10% of AA cases(13, 15). BCORL1, also known as BCL6 corepressor-like protein, shares many features with BCOR. Most mutations result in stop codon gains, frame-shift insertions or deletions, splicing errors, and gene loss, resulting in the loss of BCOR function(39). They also result in a reduction in mRNA levels suspected to be due to activation of nonsense-mediated mRNA decay pathway(39). PRC1.1 complex contains BCOR and is associated with hematopoietic regulation by repressing myeloid regulatory genes and instead committing the progenitors towards lymphopoiesis; it is this loss in function that results in a selective disadvantage in B- and T-cell lineages(40).

AA patients carry a disproportionately higher frequency of BCOR and BCORL1(13, 41), suggesting that these mutations are preferentially selected in the failing marrow environment and, in fact, are not age-related somatic mutations. The risk of transformation to MDS/AML in AA patients over a period of 10 years is estimated to be between 10–15%(12, 13) overall, but subset analysis shows that patients with BCOR and BCORL1 (favorable risk mutations) harbor a smaller risk of transformation when compared to driver mutations such as ASXL1, RUNX1, and splicing factor mutations (unfavorable risk)(23).

Whole exome sequencing has also identified BCOR and BCORL1 mutations in other hematological diseases, though typically at lower frequency than is seen in AA, such as MDS (4.2%), AML (3.8%), secondary AML (3.5%), CMML (7.4%) (42, 43).

Structural mutations in HLA and 6pLOH loss

HLA allele lacking clones may represent an immune escape of HSCs from T-cell directed immune attack and destruction in AA. While the exact pathophysiological association of HLA gene variations and immune AA remains elusive, the mechanisms suspected to be responsible are autoreactive T cell destruction of the HSCs and progenitor cells, hematopoietic micro-environmental dysfunction, immune balance dysregulation and changes in cell cycle check point gene levels. (44). Specific HLA alleles have also been associated with either increased risk of developing AA or risk of malignant clonal evolution in AA patients(14). A meta-analysis by Deng and associates(45) demonstrated an increased risk of immune AA with HLA-A and HLA-DRB1 but a protective effect of other HLA-DRB alleles. HLA-DRB1* 1501 was reported to be predictive of IST response.

In a recently published retrospective analysis of 544 immune mediated AA patients, Zaimoku and colleagues(46) reported HLA class I loss in 92 (22%) of the 412 tested subjects. Recurrently inactivated class I alleles with 393 somatic HLA gene mutations and 40 instances of loss of heterozygosity were reported. HLA-B*14:02, followed by HLA-B*02:01, HLA-B*40:02, HLA-B*08:01 and HLA-B*07:02 were most frequently mutated resulting in HLA loss. High-risk clonal evolution was associated with HLA-loss, genotype HLA-B*14:02 and older age. HLA-B*40:02 loss was also correlated with higher blood counts, while loss of HLA-B*07:02 and HLA-B*40:01 correlated to late-onset AA.

Antigen presentation by hematopoietic stem/progenitor cells to cytotoxic T cells via the HLA-B allele likely plays a critical role in the pathogenesis of immune mediated AA as is suggested by the high prevalence of leukocytes lacking HLA-B*40:02 due to 6pLOH, structural gene mutations, or both(46). When looking at HLA class II alleles, it has been shown that patients with AA exhibit low structural divergence which may contribute to T-cell dysfunction and autoreactivity. In AA patients, lower evolutional diversity in HLA II has been associated with increased risk of clonal evolution, likely due to reduced immune surveillance(47).

Clonal T-cells in AA

Somatic mutations in the JAK-STAT and MAPK pathways are frequently observed in patients with immune AA. Lundgren et al.(48) found a significant increase in JAK-STAT mutations in either CD4+ or CD8+ T-cells, present in in 75% (18/24) of the AA cohort vs. only 40% [8/20] in healthy individuals. CD8+ T cells harbored a greater burden of mutations when compared to CD4+ T cells, and somatic mutations may be a mechanism that contribute to T-cell activation. In AA, CD8+ specific mutations were restricted to a common T cell clone indicating a post thymic emergence(48).

When CD3- bone marrow cells of a healthy individual are exposed to the CD8+ cytotoxic T cells of an untreated AA patient in-vitro, apoptosis is observed along with an inhibition to CD34+ colony formation(49, 50). Several studies examining clonal selection and diversity have demonstrated large clonal expansions of activated CD8+ cytotoxic T cells in both blood and bone marrow samples, particularly in cases following strong antigenic stimuli such as CMV or EBV infection(51–54). AA often has a restricted CD8+ and CD4+ cytotoxic cell receptor repertoire (TCR) signifying an oligoclonal T cell expansion which supports the hypothesis of an antigenic stimulus. The restricted pattern on TCR is predictive of response to IST, and this response often results in restoration of T-cell variability(55, 56). Restricted TCR variability supports the immune basis of AA pathophysiology.

Large granular lymphocytosis (LGL) is a clonal expansion of T cells that results from a somatic STAT3/STAT5b mutations in approximately 40% of cases, with the cause of the remaining unclear but often suspected to be reactive(57). The presence of STAT3 mutations, however, is not pathognomonic for LGL. In a small percentage of patients with AA (7%) and MDS (2.5%), STAT3 has been reported independently of clonal LGL expansion(58). LGL results in an increased count of large granular lymphocytes, that contain azurophilic granules with acid hydrolases. Often cytopenias are seen, with single, or pan-lineage effects. LGL can co-exist with multiple bone marrow failure disorders including MDS, pure red cell aplasia (PRCA) or immune AA(57, 59). CD8+ cells are increased in number relative to CD4+ cells and clonality noted on the V-beta families when TCR is tested.

Therapeutic implications of clonal evolution

AA was once a universally fatal disease, however, advances in BMT, as well as the introduction of horse ATG (hATG) in the 1980s, later with the combination of cyclosporine, as a successful IST regimen has changed the treatment landscape. Current overall response rates to IST range have reached over 70% with the addition of EPAG(15, 60). However, long term risk of clonal evolution remains a concern in patients who receive IST. To date, the addition of EPAG has not significantly increased risk of high-risk evolution events in two reported studies in the upfront setting, though follow-up remains short(15, 60). In the NIH EPAG treated cohort, high-risk clonal evolution (defined as morphological neoplasia, or isolated chromosome 7 or complex karyotype) was 5.7% which was not higher than those treated historically with hATG and cyclosporine without EPAG. However, evolution was reported earlier in the IST and EPAG treated cohort and was usually evident by 6 months(60). In the refractory SAA setting, clonal evolution was reported to be 18% with EPAG monotherapy and these patients may warrant closer follow up(61). EPAG has not been correlated with an increased presence or expansion of somatic clones compared to patients who did not receive EPAG (15, 60).

In the recently published RACE study(15), the frequency and VAF of somatic mutations was not shown to correlate with rate or depth of response. There was an observed increase in the frequency of mutations in both treatment cohorts 6 months after therapy from 30% at baseline to 55% in those who received EPAG and 66% in those who did not. New mutations were also acquired at 24 months in both groups, where there appeared to be higher numbers in the non-EPAG treat patients. However, longer term follow-up is required to fully assess the impact of this observed clonal expansion after therapy.

BMT offers a curative option for SAA compared to IST, where relapse and clonal evolution may occur long term. High morbidity and mortality have precluded its upfront use in older patients, though increasingly MUD transplant is being used in treatment naïve pediatric patients. Haploidentical transplants offer a readily available donor pool and a shorter time transplant; clinical studies are ongoing assessing its role in upfront SAA. Currently, BMT is indicated for patients who develop a high-risk karyotype after IST. One recent analysis showed OS rates of <10% in high-risk evolvers who did not undergo BMT compared to approximately 50% for those who did (23). In patients who develop isolated non-chromosome 7 abnormalities, outcomes are favorable and often these clones can be expectantly monitored. Less clear is whether the appearance of a high-risk somatic mutation, such as RUNX1 or ASXL1 warrants consideration of earlier transplant, and this is an area of ongoing study. Conversely, the appearance of BCOR/L1 or DNMT3A does not seemingly increase risk of malignant progression.

Final conclusions

The immune marrow environment likely shapes the development of clones in AA resulting in immune escape mechanisms or driving malignant transformation. Enhanced mutational profiling, in the era of deep NGS, has improved the diagnostic specificity of identifying myeloid neoplasms and thus clonality is increasingly recognized. Certain karyotypes and somatic mutations have been associated with malignant transformation to MDS/AML, whereas others have a favorable outcome. While the effect of certain chromosomal abnormalities is well described and their presence or absence currently guides therapy, data on somatic mutations is still emerging. To date, the presence of ASXL1, RUNX1, and splicing factor mutations in AA patients seem to confer a high-risk of malignant transformation to MDS/AML while BCOR/L1 and PIGA conversely confer a better prognosis and response to IST. In the future, molecular data are likely to be incorporated into the diagnostic and therapeutic algorithms for AA.

Figure.

A. Acquired aplastic anemia is hypothesized to result from an immune mediated cytotoxic T-cell destruction of the HSCs. It is possible for the HSCs to have pre-existing mutations even prior to the immune mediated attack. The immediate consequence of the event is reduced number of HSCs (hypocellularity). As the HSCs compartment recovers, there is the possibility of acquisition and persistence of new somatic mutations at a very small VAF and otherwise no hematological consequence or clonal expansion. It is suspected that the clonal expansion of the mutated cells over the normal HSCs may be due to the better proliferation potential and survival advantage favoring these clone(s) in the recovering marrow. Alternatively, if during the recovery phase somatic mutations with neoplastic potential are acquired this can result in clonal expansion and clinically relevant effects. Another possible outcome is via immune escape mechanism, whereby the acquisition and expansion of somatic mutations such as PIGA mutation allows for recovery of the HSCs and thus avoids the immune attack. B. Possible outcomes of with HSC recovery and clonal expansion after the initial immune mediated event. Top: HSC recovery with a small clone that remains stable over time and produces no obvious clinically relevant effects. Middle: HSC recovery with acquired (new) and or preexisting (old) mutations that result in clonal expansion secondary to a superior proliferative potential over time. It is possible for the driver mutation(s) to be acquired at the early stages of the immune attack or during the later course, consistent with the Knudson “two-hit” hypothesis or a second mutation in a subclone. Bottom: HSC recovery with immune escape mechanism and clonal evolution where the expanded clones provide a competitive advantage for survival such as the development of PNH due to PIGA mutation which co-exists with normal HSCs. C. Frequency of somatic mutations in CHIP or myeloid cancer genes as observed in immune mediated acquired aplastic anemia.

Abbreviations:

CH

Clonal hematopoiesis

BMF

Bone marrow failure

IST

Immunosuppressive therapy

EPAG

Eltrombopag

BMT

Bone marrow transplant

PNH

Paroxysmal nocturnal hematuria