The complex pathophysiology of acquired aplastic anaemia

Abstract

Immune-mediated destruction of haematopoietic stem/progenitor cells (HSPCs) plays a central role in the pathophysiology of acquired aplastic anaemia (aAA). Dysregulated CD8⁺ cytotoxic T cells, CD4⁺ T cells including T helper type 1 (Th1), Th2, regulatory T cells and Th17 cells, natural killer (NK) cells and NK T cells, along with the abnormal production of cytokines including interferon (IFN)-γ, tumour necrosis factor (TNF)-α and transforming growth factor (TGF)-β, induce apoptosis of HSPCs, constituting a consistent and defining feature of severe aAA. Alterations in the polymorphisms of TGF-β, IFN-γ and TNF-α genes, as well as certain human leucocyte antigen (HLA) alleles, may account for the propensity to immune-mediated killing of HSPCs and/or ineffective haematopoiesis. Although the inciting autoantigens remain elusive, autoantibodies are often detected in the serum. In addition, recent studies provide genetic and molecular evidence that intrinsic and/or secondary deficits in HSPCs and bone marrow mesenchymal stem cells may underlie the development of bone marrow failure.

Introduction

Aplastic anaemia (AA) is a bone marrow failure syndrome characterized by bone marrow aplasia and peripheral blood pancytopenia. Patients with AA often present with symptoms of anaemia, purpura or haemorrhage and less frequently infection, leading to medical evaluation 1,2. Most cases of AA are acquired and idiopathic. Although the pathophysiology of AA is not completely understood, the fact that up to 80% of the patients with AA will respond to immunosuppressive therapy (IST) with anti-thymocyte globulin (ATG) and cyclosporin (CsA) implies an underlying immune pathophysiology 2,3. Although long-term event-free survival (EFS) can be achieved with IST in many patients, the risk of relapse and clonal evolution to myelodysplastic syndrome (MDS) or myeloid leukaemia remains high. Matched related haematopoietic stem cell transplantation (HSCT), when feasible, is the preferred definitive treatment for severe AA in children and young adults 1.

A large amount of laboratory and clinical data suggest that immune-mediated suppression of haematopoiesis plays an essential role in the pathogenesis of acquired AA (aAA). However, recent studies provide evidence that haematopoietic stem/progenitor cells (HSPCs) and bone marrow mesenchymal stem cells (BM-MSCs) from patients with aAA carry intrinsic deficits that contribute to their vulnerability in developing bone marrow failure. In this review, we will discuss the recent advances in the pathophysiology of aAA, focusing primarily on dysregulated immune responses.

Influence of immunogenetics

The association between HLA regions and aAA has been suggested in a number of studies. However, these studies are often limited by the following factors: (1) small sample size due to the rarity of the disease; (2) different distribution of patient age (paediatric versus adult aAA patients) in each study; and (3) linkage disequilibrium differs between HLA alleles in different populations, thus making it difficult to draw definitive conclusions from observations based on specific ethnic and age groups. Studies with larger patient populations from diverse ethnic backgrounds are warranted.

Nevertheless, the existing data support the potential influence of HLA polymorphism on the pathophysiology of aAA (summarized in Table1). HLA alleles that are associated with increased 4–9 or decreased 7,10 susceptibility to aAA have been reported. In addition, certain HLA alleles appear to play a role in predicting responses to IST 4,11. Conflicting reports exist regarding the correlation between human leucocyte antigen (HLA)-DRB1*1501 and the presence of a paroxysmal nocturnal haemoglobinuria (PNH) clone and a good response to IST 11–13.

Table 1.

Influence of immunogenetics in aplastic anaemia (AA)

HLA allelles increasing susceptibility to aAA	HLA allelles decreasing susceptibility to aAA	HLA alleles predicting response to IST
HLA-DRB1*1501 4,5	HLA-DRB1*03:017	HLA-DRB1*04 11
HLA-DRB1*0405 6	HLA-DRB1*11:017	†HLA-DRB1*1501 11–13
HLA-DRB1*09:01 7	HLA-DRB1*03 10	HLA-DQB1*0401 11
HLA-DRB1*07 9	HLA-B*51:01 7	HLA-DQB1*0602 11
HLA-B14 8		HLA-DRB1*1501-DQA1*0102-DQB1*0602 haplotype 4
HLA-B *48:01 7

^†Conflicting data exist. HLA = human leucocyte antigen; aAA = acquired aplastic anaemia; IST = immunosuppressive therapy.

How do specific HLA alleles confer susceptibility to AA? Nakao et al. isolated a CD4⁺ Vbeta21⁺ T cell clone in a patient with aAA that was capable of killing haematopoietic cells in an HLA-DRB1*0405-restricted pattern, indicating that cytotoxic T cells may contribute to the pathogenesis of aAA 6. This observation suggests that certain HLA alleles may play a role in the activation of autoreactive T cell clones in patients with aAA. Furthermore, the protective effects of given HLA molecules has been suggested to be due to the insufficient generation of autoreactive regulatory T cells (T_regs) that suppress autoimmunity 14. This, in combination with the observation that T_reg frequency corresponds with disease severity and response in aAA 15, argues that some HLA molecules may be associated with failure to protect, rather than with active induction of an immune attack in aAA.

Dysregulated T cell responses in aAA

Genome-wide transcriptional analysis of T cells from aAA patients has revealed a large number of dysregulated genes in patients' CD4⁺ and CD8⁺ T cells 16. A combination of abnormal expansion of Th1, Th2, and Th17 cells and a decreased or skewed T_reg immunophenotype and function constitutes a consistent and defining feature of severe aAA 17.

CD8⁺ T cells

The critical role of CD8⁺ T cells in the pathogenesis of aAA is supported by a variety of clinical and experimental evidence. Significantly higher numbers of CX3C chemokine receptor 1 (CX3CR1)-expressing CD3⁺ T cells were identified in the bone marrow of patients with severe aAA, suggesting a site-directed recruitment of T cells 18. The early observation that marrow and peripheral blood lymphocytes from patients with aAA are able to suppress haematopoiesis in vitro indicate that T cells play a major role in the pathophysiology of aAA. Subsequently, activated circulating CD8⁺ T cells were identified as the lymphocyte subset that inhibited haematopoiesis in aAA patients 19. However, neither the immune response nor the nature of the inciting antigen(s) has been well characterized.

Systematic analysis of the T cell receptor (TCR) repertoire by (i) TCR V-beta flow cytometry, (ii) complementary-determining region 3 (CDR3)-specific amplification and sequencing and (iii) spectratyping to detect skewing of TCR CDR3 size allows identification of molecular clonotypes and quantification of clonal expansion in patients with aAA. CD8⁺ cytotoxic T cells (CTLs) with highly restricted TCR diversity (oligoclonal T cells) have been detected in aAA 2. The predominance of certain V-beta families, including V-beta 17, appears to support an autoimmune process directed against common antigen(s), but the high diversity of the antigen-binding sites among the analysed aAA patients, identified at the single-cell level, argues for the predominance of private and not common/public inciting epitopes 20. Conversely, a report that nearly all HLA-matched aAA patients share almost identical pathogenetic clonotypes, with homology greater than 98% for the entire β chain, strongly supports public immune responses triggered by similar or identical antigens, presented within the same HLA molecules 21.

The detection of immunodominant clonotypes with a significantly skewed TCR V-beta CDR3 repertoire in aAA suggests antigen-driven expansion of T cells. A correlation between the frequency of the clonotypes and haematological response to IST supports the disease specificity of these clonotypes, indicating that they may be used as markers to assess disease activity, treatment response and relapse 22,23. It has been reported that in addition to the decline in the frequency of the immunodominant clones, restoration of TCR variability can be observed in aAA patients who respond to IST 12,24. At the time of disease relapse, the original, putatively pathogenetic dominant clones reappear and are sometimes accompanied by new clones, consistent with epitope spreading of the immune response. Occasionally, a clone persists in remission, suggesting T cell tolerance 2,12,23.

Why T cells are activated in aAA remains to be clarified. Over-representation of certain HLA alleles among aAA patients, as outlined above, suggests a role for antigen recognition. Aberrant expression of TCR signalling-related genes may contribute to T cell dysfunction in aAA. Li et al. reported that expression levels of the CD3γ, CD3δ, CD3ε and CD3ζ genes were increased significantly in patients with aAA compared to healthy controls. Up-regulation of CD3 gene expression in aAA suggests that T cells may be in sustained signalling stimulation in the periphery that leads to inappropriate T cell activation 25. Contradictory to this report, it was found that expression of the CD3 ζ-chain was decreased in most patients with aAA 26. Further studies are needed to explore the underlying mechanisms of aberrant CD3 gene expression. Nevertheless, these observations suggest dysregulated T cell activation pathways in aAA.

CD4⁺ T helper cells

Abnormality in the number and/or function of CD4⁺ cells, including interferon (IFN)-γ-producing CD4⁺ T cells (Th1 cells), interleukin (IL)-4-producing CD4⁺ T cells (Th2 cells), T_regs and interleukin-17 (IL-17)-producing CD4⁺ T cells (Th17 cells) has been reported in patients with aAA, suggesting their potential roles in the pathogenesis of the disease. Giannakoulas et al. reported that untreated or refractory aAA patients had a significantly higher proportion of unstimulated Th1 cells that produced IFN-γ and IL-2, whereas Th2 cells did not differ from that of controls, resulting in a shift of the IFN-γ/IL-4 ratio towards a type-1 response. Patients in remission also had an increased proportion of Th1 cells, with a parallel rise of Th2 cells and normal IFN-γ/IL-4 ratio 27. Spectratyping and high-throughput sequencing reveals that, similar to CD8⁺ T cells, the clonality of Th1 cells in aAA is restricted 17,21,28, again suggesting an antigen-driven expansion of Th1 cells. Functionally, these dominant CD4⁺ Th1 clones in aAA secrete IFN-γ and tumour necrosis factor (TNF)-α and are capable of lysing autologous CD34⁺ cells and inhibiting their haematopoietic colony formation 28.

T_regs

T_regs are CD4⁺CD25⁺forkhead box protein 3 (FoxP3)⁺ cells that play fundamental roles in autoimmunity. Human FoxP3⁺CD4⁺ T cells are composed of three phenotypically and functionally distinct subpopulations: CD45RA⁺FoxP3^low resting T_reg cells (rT_reg cells) and CD45RA⁻FoxP3^high activated T_reg cells (aT_reg cells), both of which are suppressive in vitro, and cytokine-secreting CD45RA⁻FoxP3^low non-suppressive T cells 29. The proportions and functions of these T_reg subpopulations vary in different autoimmune disorders. The number of T_regs correlates with disease severity in patients with aAA 15,17. More detailed evaluation of distinct T_reg subpopulations in aAA reveals a skewing of T_reg subsets. Numbers of both activated and resting T_regs are reduced in patients with aAA. In contrast, cytokine-secreting non-T_regs are increased. Functionally, T_regs from aAA patients display intrinsic impairment. They have impaired migratory ability because of lower CXCR4 expression and have decreased capacity in suppressing normal effector T cell functions, including IFN-γ production 17,30. Moreover, patients with higher numbers of T_regs are more likely to respond to IST 17.

Th17 cells

Th17 cells are CD4⁺ T cells that secrete IL-17A, a cytokine that co-ordinates tissue inflammation by inducing chemokines (such as CXCL8, CXCL6 and CXCL1), growth factors [such as granulocyte-colony stimulating factor (G-CSF), granulocyte–macrophage-colony stimulating factor (GM-CSF) and IL-6] and adhesion molecules [such as intercellular adhesion molecule-1 (ICAM-1)], leading to augmented neutrophil accumulation as well as to granulopoeisis. The association between IL-17 and autoimmune disorders such as inflammatory bowel disease, rheumatoid arthritis and lupus is now well recognized 31. However, the role of Th17 cells in aAA has not been well established. IL-17 does not seem to be elevated in the plasma of most patients with aAA. Indeed, IL-17 was undetectable in the plasma of patients with severe aAA in all but one published study. 17,32. In some studies, Th17 cells expanded in aAA and correlated with disease activity. Furthermore, the expansion of Th17 cell population correlated with the depletion of natural T_regs in the blood of aAA patients, indicating a reciprocal relationship between Th17 cells and CD4⁺CD25^highFoxP3⁺ T_regs and underlying the autoimmune process in aAA 17,33. In a murine model of immune-mediated bone marrow failure, early depletion of Th17 cells using anti-IL-17 antibody increased the number of T_regs, decreased IFN-γ levels and reduced the severity of bone marrow failure, suggesting that the Th17 immune response may contribute early in the phase of aAA development. It was hypothesized that the immune system is polarized through a Th1/Th17 response during the induction of bone marrow failure, with a T_reg deficiency leading to the increased autoreactive T cell activity and the development of clinical aAA 33. However, in another study, Kordasti et al. reported that there was no difference in the frequencies of Th17 cells between patients with aAA and healthy controls 34. This discrepancy may be the result of different percentages of moderate aAA patients who display higher numbers of Th17 cells in the former cohort 33.

Collectively, the currently available data suggest that a combination of expansion of Th1, Th2 and possibly Th17 and a decreased or skewed T_reg immunophenotype and function probably contribute to the pathogenesis of aAA, particularly severe and very severe aAA. The precise sequence of events that lead to haematopoietic failure is not clear, but the current evidence suggests that expansion of Th1 cells is likely to be antigen-driven; Th17 cells may modulate Th1 cells directly or indirectly; and that the presence of dysfunctional T_regs may aggravate this autoimmune response 17,33.

Putative autoantigens in aAA

As reviewed above, it is widely accepted that T cell-mediated cellular immunity plays a critical role in the pathogenesis of aAA. However, the putative autoantigens that induce the aberrant immune responses against haematopoietic stem/progenitor cells (HSPCs) remain unknown. A variety of autoantibodies have been detected in serum of patients with aAA. These have included anti-moesin, diazepam-binding inhibitor-related protein 1 (DRS1), kinectin, post-meiotic segregation increased 1 (PMS1), heterogeneous nuclear ribonucleoprotein (hnRNP) K, CLIC1, HSPB11 and RPS27 antibodies. Furthermore, the presence of anti-hnRNP K, anti-moesin, anti-DRS1, anti-CL1C1 and anti-RPS27 autoantibodies is correlated significantly with a good response to IST 35–38. There is some in-vitro evidence that these putative autoantigens may elicit immune attack against HSPCs 36,39. For example, reactive CD8⁺ cytotoxic T cells against kinectin generated in vitro were capable of suppressing granulocyte–macrophage colony-forming units (CFU-GMs) in an HLA class I-restricted fashion. However, anti-kinectin T cells were not identified in aAA patients 35. A CD4⁺ T cell epitope derived from DRS1 protein successfully induced cytotoxic T cells obtained from one patient with aAA. An increased frequency of T cell precursors specific to the DRS-1 peptide were detected in two patients 36. However, the lack of specificity to HSPCs limits the clinical relevance of these autoantigens and autoantibodies.

Innate immunity

Although the majority of the studies have focused upon T and B cells, there is emerging evidence indicating that dysfunctional innate immunity may also play a role in the pathogenesis of aAA.

Conflicting observations exist in interpreting the role of NK cells in aAA. Some studies show that NK numbers and cytolytic activity are impaired in aAA and recovery of NK cytotoxicity correlates with haematopoietic recovery following IST 40. The deficiency in NK cytolytic activity may be secondary to intrinsic perforin gene mutation or the result of suppression by autologous granulocytes in patients with aAA 41. However, in one study of paediatric aAA patients the NK cell frequency did not correspond with disease severity or response, which raises the question of whether NK cell deficiency is the result, rather than the trigger, of bone marrow failure15.

Abnormal expression of stress-inducible natural killer group 2, member D (NKG2D) ligands, such as UL16-binding protein (ULBP) ULBP1, ULBP2 and ULBP3, and major histocompatibility complex class I chain-related molecules A (MICA) on granulocytes and bone marrow cells has been reported in aAA 42. The expression level of NKG2D ligands on granulocytes correlates positively with bone marrow failure and a favourable response to IST in aAA patients. In-vitro studies reveal that aAA haematopoietic progenitor cells with abnormal NKG2D ligand expression are damaged by autologous lymphocytes bearing NKG2D, including NK, NK T, CD8⁺αβ T, γδ T cells and a small subset of CD4⁺ T cells, consistent with the findings in PNH 42,43. Altogether, these findings suggest that NKG2D-mediated immunity, which drives NK, NK T and T cell activation, is involved, at least partially, in the pathogenesis of aAA. Detection of NKG2D ligands on haematopoietic cells may be valuable in the early diagnosis of immune-mediated marrow damage and potentially in predicting response to IST.

Myelosuppressive cytokines

The self-reactive T cells in patients with aAA secrete proinflammatory cytokines such as IFN-γ and TNF-α, resulting in elevated cytokine levels in the bone marrow and peripheral blood of aAA patients 44–46. IFN-γ and TNF-α reduces colony formation of human haematopoietic progenitor cells in vitro by inducing apoptosis of CD34⁺ cells through the Fas–Fas ligand pathway 47,48 and/or the TNF-related apoptosis-inducing ligand (TRAIL) pathway 49. Furthermore, IFN-γ leads to AA by disrupting the generation of common myeloid progenitors and lineage differentiation in the experimental murine model 50. The presence of intracellular IFN-γ in T cells in the blood and bone marrow of aAA patients may predict a response to IST, but also the onset of relapse 46. In contrast, intracellular expression of TNF-α in marrow T cells is associated with unfavourable clinical outcome 51.

Various groups have investigated polymorphisms in these myelosuppressive cytokines in an attempt to identify a genetic predisposition to aAA or degree of response to IST. The TNF-α gene −308 promoter/enhancer polymorphism, and specifically the TNF2 allele (−308A), was over-represented in groups of Chinese and German aAA patients and was associated with an improved response to IST 52,53. However, contradictory observations exist; e.g. an association was not evident in a study of Korean aAA patients 54. These discrepancies between studies may be attributed to the small numbers of evaluable patients and perhaps to differences between the ethnic groups evaluated.

Several IFN-γ gene polymorphisms, including the IFN-γ −2353 T allele and TCA haplotype as well as hypersecretory genotype T/T at position −874 of the IFN-γ gene, were found to be over-represented in aAA. In addition, this relevant IFN-γ −2353 T allele and TCA haplotype were related to the response of IST in a Korean study 54,55. T-bet belongs to the T-box family of transcription factors and is the key regulator of Th1 development and function 56. In patients with aAA, the increased IFN-γ levels were found to be the result of active transcription of the IFN-γ gene by T-bet 57.

Transforming growth factor (TGF)-β1 is another cytokine with multi-functional effects that plays a role in haematopoiesis. The frequency of genotypes associated with high production of TGF-β1, and in particular the −509 TT genotype, was increased in patients with aAA 53,55. In contrast, Rizzo et al. demonstrated lower levels of TGF-β1 in the serum and in bone marrow stromal cell cultures of aAA patients, hypothesizing that accessory cells in the bone marrow compartment down-regulate TGF-β1 expression to allow haematopoietic stem cell cycling to counteract hypoplasia 58. Other polymorphisms of TGF-β1, such as −590 C/T rs1800469 as well as the P10L C/T rs1800470, have been reported to play no role in the susceptibility to aAA. However, they may be associated with higher responses to IST 54.

Haematopoietic stem/progenitor cells

Although immune-mediated destruction of bone marrow haematopoietic stem/progenitor cells (HSPCs) is a well-known underlying mechanism of aAA pathophysiology, recent studies reveal that the HSPCs in subgroups of patients with AA carry intrinsic qualitative deficits.

Significant telomere shortening in leucocytes has been reported in subpopulations of patients with aAA, especially those who do not respond to IST 59,60. Mutations in telomerase complex genes such as telomerase RNA component (TERC) and telomerase reverse transcriptase (TERT) lead to defects in the maintenance of telomere length, resulting in deficient haematopoietic survival and proliferative capacity, and ultimately in a reduced haematopoietic stem cell pool 61–63. These patients, when exposed to an environmental insult that induces immune-mediated haematopoietic stem cell destruction, may be more susceptible in developing bone marrow failure and with suboptimal responses to IST.

However, the majority of AA patients with short telomeres lack known genetic mutations responsible for shorter telomeres, suggesting the involvement of other genes and/or environmental factors 64. Accelerated telomere shortening in subsets of untreated or refractory AA patients has been implicated to be a result of increased HSPC proliferation, similar to what has been reported in patients following allogeneic HSCT 59,65,66. Therefore, instead of underlying the genetic aetiology of the disease, in subpopulations of patients with AA short telomeres may be the consequence of restricted clonal haematopoiesis and ‘regenerative stress' leading to chromosomal instability in HSPCs 67,68. Furthermore, it may also reflect DNA damage to HSPCs following previous exposure to reactive oxidative stress or other environmental factors. Nevertheless, there is a significant correlation between telomere length and persistent cytopenias after IST. Recent studies demonstrate that shorter telomere length is associated with a greater risk of relapse, clonal evolution and monosomy 7 as well as inferior overall survival 64,68.

Regardless of the aetiology of short telomeres in patients with aAA, short and dysfunctional telomeres not only restrict the proliferation of normal HSPCs, but also produce chromosomal instability and predispose to malignant transformation. Accelerated telomere attrition preceded by aneuploidy and clonal evolution to monosomy 7 suggests that critically short telomere length may service as a biomarker for clonal evolution in patients with aAA. Therapeutic up-regulation of telomerase by androgen or other pharmacological agent might reduce the risk of clonal evolution 68.

In addition, CD34⁺ HSPCs from patients with aAA also display marked down-regulation of cell cycle ‘checkpoint' genes such as cyclin-dependent kinase 6 (CDK6), CDK2, cyclins E and A, Fanconi anaemia complementation group (FANCG), c-myb and c-myc. These may provide additional mechanisms explaining (1) the inability of remaining HSPCs to replicate competently and ultimately compensate for immune-mediated destruction and (2) the development of premalignant or aneuploid cells in AA patients, who remain susceptible to clonal evolution to myelodysplasia or myeloid leukaemia even years after achieving haematological recovery with IST 69,70. Alternatively, it is possible that these cell cycle defects may be secondary to telomere dysfunction in aAA.

Furthermore, a recent study revealed that acquired somatic mutations in myeloid malignancy-related genes, including ASXL1, DNMT3A, TET2 and BCOR, are present in a proportion of patients with aAA and predict high risk of malignant transformation to MDS/AML. Interestingly, the majority of these mutations are genes involved in epigenetic regulation of DNA transcription, suggesting potential association and co-operation between mutations in epigenetic regulators and immune-mediated bone marrow failure. The presence of these mutated clones could indicate clonal myelopoiesis, such as that in MDS, selective adaptation in the context of IST or the normal ageing process. Large prospective studies with sequential analysis of somatic mutations are indicated to determine more accurately the impact and dynamics of mutations in predicting the risk of evolution to MDS and AML in aAA 60. Findings from these studies may have implications regarding treatment decisions, arguing for earlier haematopoietic stem cell transplantation rather than continued exposure to IST in subgroups of aAA patients harbouring these mutations.

BM-MSCs

BM-MSCs are key precursor cells of the marrow microenvironment that play important roles in maintaining long-term haematopoiesis. BM-MSCs differentiate into a variety of stromal cells that constitute the haematopoietic stem cell niche. BM-MSCs and differentiated stromal cells support haematopoiesis and regulate overall immune cell function to maintain haematopoietic and immune homeostasis through the production and secretion of cytokines, chemokines and extracellular matrix molecules. Given that MSCs possess remarkable immune-suppressive properties on T cells, NK cells and antigen-presenting cells (APCs) 71, it is plausible to reason that dysfunction of BM-MSC may contribute to the pathogenesis of AA.

The existing literature is small, and conflicting results have been reported regarding the phenotype and function of BM-MSCs in patients with AA. Some studies demonstrated that MSCs from aplastic bone marrow harboured intrinsic and/or secondary defects, including poor proliferation, reduced clonogenic potential, increased apoptosis, aberrant differentiation and inadequate suppression of T cell activation, TNF-α and IFN-γ production in vitro 72–76. Interestingly, the deficiency in BM-MSCs to down-regulate T cell priming, proliferation and cytokine release persists indefinitely after immunosuppressive therapy, but may be restored after bone marrow transplant (BMT) 76. Some other studies revealed normal immunophenotype and immunosuppressive properties of BM-MSCs from patients with AA 77,78. Such discrepancy between these studies may reflect differences in pathogenesis between (1) adult AA and paediatric AA and/or (2) moderate–severe AA and very severe AA. Future studies involving larger numbers of AA patients are necessary to determine whether or not age at diagnosis and the severity of disease affect the function of BM-MSCs in AA.

Nevertheless, gene expression profiling of BM-MSCs from aAA patients revealed a marked difference in the expression of a large number of genes implicated in cell cycle, cell division, proliferation, chemotaxis, adipogenesis-cytokine signalling and haematopoietic cell lineage differentiation 79–81, strongly suggesting altered cellular function. The expression of GATA binding protein 2 (GATA2), a transcription factor critically required in the genesis and function of HSPCs, is decreased not only in HSPCs but also BM-MSCs in patients with AA due to unknown mechanisms 69,80,82. Down-regulation of GATA2 in BM-MSCs compromises colony-forming capacity of human HSPCs and accelerates adipocyte differentiation in vitro 83. Such findings suggest indirectly that decreased GATA2 expression in HSPCs and BM-MSCs may contribute to a reduced HSPC pool as well as fatty marrow change, which are characteristic features of AA. Although the molecular mechanism is unknown, decreased expression of cell adhesion molecules such as LAMB1, CD44 and FBN2 in BM-MSCs may be involved in impaired HSPC support when GATA2 is down-regulated 83. Interestingly, although implicated in the pathogenesis of AA, pro-inflammatory cytokines including TGF-β, IFN-γ, TNF-α, IL-6, IL-17A and IL-1β failed to accelerate adipocyte differentiation or decrease GATA2 expression in BM-MSCs in vitro 83, suggesting that these cytokines promote AA through other cellular populations.

The expression level of CXCL12 was also found to be decreased significantly in BM-MSCs from children with severe AA. Functionally, knocking down CXCL12 in BM-MSCs from healthy donors led to decreased survival and differentiation potential of these cells 81, indicating that down-regulation of CXCL12 expression in BM-MSCs may be associated with alterations in the BM microenvironment in AA. Given that interactions between CXCL12 and CXCR4, its main receptor expressed by HSPCs, are crucial for regulation of a number of signalling pathways involving HSPC survival, proliferation, adhesion and migration, reduced CXCL12 in BM-MSCs may contribute to defective haematopoiesis in patients with AA.

In addition to the abnormalities of MSCs revealed by in-vitro studies, co-transplantation of allogeneic BM- or cord blood-derived MSCs and HSPCs enhances haematopoietic engraftment, reduces graft-versus-host disease (GVHD) and improves stromal function in patients with AA 84–86. These reports not only confirm the immune modulatory properties of MSCs, but also provide indirect in-vivo evidence that BM-MSCs may contribute to the pathogenesis of AA.

Treatment considerations

Given the underlying immune-mediated quantitative and intrinsic qualitative deficits of HSPCs in aAA, the treatment goal for aAA is to stop the self-directed haematopoietic autoimmune attack or to replace the patients with new healthy HSPCs. HLA-matched sibling donor haematopoietic cell transplantation (HCT) is the first line of treatment for children and young adults with severe aAA. However, for greater than 70% of the patients without a matched sibling donor, IST with CsA and horse ATG is the treatment of choice 87. Clearer understanding of the immune mechanisms of aAA may help in discovering a newer generation of immune suppressing/modulating agents. With high-resolution HLA typing, improved supportive care and more refined conditioning regimens, the outcome of alternative donor HCT continues to improve 87,88. Early HCT with alternative donor grafts is now being recommended for many patients who have failed an initial course of IST 87,89,90. By referring earlier to transplant those patients who are likely to fail IST, identification of prognostic factors for response to IST may further improve the outcome of alternative donor HCT.

Conclusions

Aberrant immune regulation in aAA involving CD8⁺T cells, CD4⁺ Th1 cells, T_regs and innate immune cells leads to destruction of HSPCs (Fig. 1). The inciting autoantigens remain to be clarified. The existing data argue for common versus private antigens on HSCPs that trigger abnormal activation of immune responses. The intrinsic deficits in HSPCs and bone marrow niche contribute further to the development of bone marrow failure. Although treatments with IST or HCT have improved the outlook for patients with aAA, better understanding of the underlying mechanisms are essential for developing more targeted therapies in selected patients.

Fig 1.

Immune-mediated destruction of haematopoietic stem/progenitor cells. Following exposure to environmental insults, patients with acquired aplastic anaemia (aAA) develop aberrant immune responses that include oligoclonal expansion of T cells and abnormal production of myelosuppressive cytokines, including interferon (IFN)-γ and tumour necrosis factor (TNF)-α, leading to apoptosis of haematopoietic stem/progenitor cells (HSPCs). Deficient regulatory T cells (T_reg), natural killer (NK) cells and mesenchymal stem cells (MSC) fail to suppress the dysregulated immune responses. Intrinsic deficits in HSPCs contribute further to the development of bone marrow failure. *Conflicting data exist.

Disclosure

The authors have declared that no competing interests exist.