Concise Review: The Bone Marrow Niche as a Target of Graft Versus Host Disease


  • Malte von Bonin,

    1. Medizinische Klinik und Poliklinik 1, Universitätsklinikum Carl-Gustav Carus, Medizinische Fakultät der Technischen Universität Dresden, Dresden, Germany
    2. German Cancer Consortium (DKTK), Heidelberg, Germany
    3. German Cancer Research Center (DKFZ), Heidelberg, Germany
    Search for more papers by this author
  • Martin Bornhäuser

    Corresponding author
    1. Medizinische Klinik und Poliklinik 1, Universitätsklinikum Carl-Gustav Carus, Medizinische Fakultät der Technischen Universität Dresden, Dresden, Germany
    • Correspondence: Martin Bornhäuser, M.D., Medizinische Klinik und Poliklinik 1, Universitätsklinikum Carl-Gustav Carus, Medizinische Fakultät der Technischen Universität Dresden, Fetscherstrasse 74, 01307 Dresden, Germany. Telephone: 49-351-458-4190; Fax: 49-351-458-5362; e-mail:

    Search for more papers by this author


Graft versus host disease (GvHD) remains a major complication after allogeneic hematopoietic stem cell transplantation and is the main cause of transplant-related mortality. In addition to visceral organ involvement, concomitant myelosuppression has been repeatedly described and the extent of cytopenia has been introduced into GvHD scoring systems. Both hematopoietic cells and cells that form the hematopoietic stem and progenitor cell niche have been identified as targets of GvHD. Although several contributing factors have been previously described, the pathophysiology of GvHD-mediated myelosuppression remains largely unclear and to date, no specific therapeutic interventions have achieved routine clinical application. This review focuses on the bone marrow as a target of GvHD, the factors that contribute to myelosuppression, and the possible therapeutic approaches. Stem Cells 2014;32:1420–1428


Hematopoietic stem cell transplantation (HSCT) has become an accepted and widely used treatment option for defined malignant and nonmalignant diseases. Initially, given the considerable transplant-related mortality (TRM), which was partly associated with myeloablative conditioning (MAC), HSCT was restricted to younger patients. However, the introduction of reduced intensity conditioning has increased the proportion of HSCT-eligible patients. Today, approximately 20,000 allogeneic HSCT procedures are performed annually worldwide [1]. Despite improvements in HSCT practices and reductions in TRM, one-third of patients will die without preceding relapse [2]. Besides infection, graft versus host disease (GvHD) remains the main cause of morbidity and mortality after allogeneic HSCT. GvHD is mediated by alloreactive donor lymphocytes that reside in the graft and target nonmalignant host tissues. However, this alloreactivity not only has negative effects but also mediates graft versus tumor (GvT) effects and supposedly facilitates engraftment. To date, investigators have not been successful in separating GVT from GvHD, and therefore in most cases the GvT effects are associated with the occurrence of GvHD. In clinical practice, achievement of a well-balanced equilibrium between GvT and GvHD still remains challenging [3].

The most important proteins on the surfaces of host cells responsible for alloreactivity are the human leukocyte antigens (HLAs), which are encoded by the major histocompatibility complex (MHC) located on chromosome 6. The classic class I HLA (A, B, and C) proteins are expressed on nearly all nucleated cells, whereas the classic class II (DRB1, DQB1, and DPB1) proteins are predominantly expressed on hematopoietic cells. Nevertheless, other proteins encoded outside of MHC (minor histocompatibility antigens or mHAs) contribute substantially to the donor T lymphocyte responses, as GvHD also occurs in patients that received transplants from HLA-matched donors [3].

Historically, acute and chronic GvHD were arbitrarily separated by the presence of symptoms before and after 100 days post-HSCT, respectively. In 2005, this time-dependent classification was replaced with a scoring system based on clinical features. The Consensus Criteria, defined during a workshop of the National Institutes of Health, not only allow the distinction of acute and chronic GvHD but also the diagnosis of overlap syndromes [4]. Acute GvHD predominantly affects the skin, gastrointestinal tract, and liver, whereas chronic GvHD is typically not organ-restricted but instead results in a rather variable complex of clinical symptoms. Also, from a pathophysiological point of view, acute and chronic GvHD decisively differ. Acute GvHD is mainly mediated by direct cell-cell-mediated cytotoxicity and is associated with a cytokine pattern that shifts toward a Th1 response (e.g., interferon [IFN] γ and interleukin [IL]-2), whereas chronic GvHD is predominantly based on soluble factors and is characterized by a Th2 response (e.g., IL-4 and IL-10). Chronic GvHD often resembles collagen vascular diseases and, autoantibody formation is frequently observed [3, 4].

Irrespective of the clinical occurrence of GvHD, HSCT itself has been shown to result in long-lasting reductions in bone marrow cellularity, persistent alterations in T- and B-cell development and functionality, and profound declines in the self-renewal capacities of hematopoietic stem cells (HSC), although these alterations are not necessarily accompanied by a reduction in the peripheral blood cell counts [5-8]. Furthermore, these pathologies also occur in patients with complete donor chimerism, indicating that engraftment and graft function are regulated by distinct pathways. Importantly, hematopoietic dysfunction has been described to be exacerbated during GvHD [8, 9]. In particular, suppressed B- and T-cell development and function have been associated with an increased susceptibility to infection and worse outcomes in patients with acute and chronic GvHD. Perturbance of the lymphoid compartment has been clearly shown to be partly mediated by the destruction of the secondary lymphoid organ microenvironment. The destruction of the thymus as primary lymphoid organ has also been linked to the perturbation of the T-cell compartment in GvHD (reviewed elsewhere [10]). T-cell depletion in the graft has been shown to result in a clear reduction in the incidence of GvHD, which is offset by an increased probability of relapse (lack of GvT) and thus results in an inferior overall outcome [11, 12]. Furthermore, T-cell depletion has been associated with high rates of engraftment failure and delayed immune reconstitution when compared with the transplantation of T-cell-replete grafts [11-14]. However, doubt is increasing with regard to the facilitation of donor T-cell-attributed immune reconstitution in clinical trials when appropriate numbers of hematopoietic stem and progenitor cells (HSPCs) are transplanted [15]. Furthermore, inferior outcomes with T-cell depletion of the graft have been mainly observed in chronic myeloid leukemia. In contrast, investigators have recently reported encouraging results with CD34+ selected hematopoietic stem cell grafts in patients with acute myeloid leukemia (AML) in remission [15, 16]. So, the underlying disease, the graft source, and the mode of T-cell depletion seem to play a relevant role. In different murine models of GvHD, T-cell depletion even resulted in accelerated B-cell reconstitution with full chimerism in the B-cell and myeloid compartments. The recipients of T-cell-depleted grafts retained a mixed chimerism only in the T-cell lineage, a phenomenon that might be critical to achieve full GvT potency [17, 18]. Indeed, incomplete T-cell chimerism might mainly occur in older patients receiving T-cell depleted grafts, as the lack of thymic function might result in almost complete dependence on homeostatic peripheral expansion of mature T cells contained in the graft.

Nonlymphoid cytopenia also frequently occurs after HSCT and has been described in the context of infections (e.g., cytomegalovirus), relapse and graft-failure, or pharmacological induction [19]. In particular, thrombocytopenia has been observed to associate with acute and chronic GvHD [20-34]. Notably, in a multivariate analysis, acute GvHD was described as one of the factors that associated significantly with the secondary failure of platelet recovery (SFPR) [26]. This association was highlighted by the concomitant reversibility of thrombocytopenia and of clinical GvHD signs under immunosuppressive therapy [34]. Furthermore, a low platelet count was proposed as an indicator of GvHD severity, and persistent thrombocytopenia has been described as an independent predictor of worse outcomes in patients with chronic and acute GvHD [20-24, 26-34]. Remarkably, patients with persistent thrombocytopenia do not die of hemorrhagic events, but rather other GvHD-associated complications, particularly infections. In patients with isolated thrombocytopenia, bone marrow aspirates and biopsies mainly show hypocellularity and decreased megakaryocytopoiesis, indicating that bone marrow dysfunction is the main reason for the low platelet count. Nevertheless, one-third of these patients will exhibit an increase in megakaryocytopoiesis consistent with peripheral platelet destruction [28] and recently, thrombopoiesis, as determined by the peripheral blood absolute immature platelet number, was reported to be markedly elevated in thrombopenic patients with chronic GvHD, thus providing evidence against myelosuppression [25]. In contrast to thrombopoiesis, myelopoiesis and erythropoiesis have not regularly been described as affected during GvHD [30, 32, 35]. However, the number of cell lineages affected by cytopenia has been associated with an inferior outcome [30].

Given these observations, GvHD-mediated nonlymphoid cytopenia might be induced both by a disturbance in HSPC maturation and the consumption of already mature hematopoietic cells. As HSPC function largely depends on the microenvironment, a perturbance of maturation could conceivably be mediated not only by targeting hematopoietic cells (direct) but also components of the niche (indirect).

The hematopoietic niche is an anatomic site where hematopoietic stem cells are sustained. According to the niche-concept previously envisioned by Schofield in 1978, this site can control the self-renewal versus the differentiation of HSPC [36]. As the niche space is considered limited with regard to available binding sites, a constant competition between incoming and egressing HSPC is assumed. Mesenchymal stromal cells (MSCs) comprise a major cellular component of the bone marrow niche. The major contribution of these cells is the production of chemokines (e.g., CXCl2/SDF-1) that mediate homing and survival. This property has been especially attributed to a subset of Nestin-positive MSC [37]. The following have been shown to be roles of MSC in the bone marrow niche: MSCs act as cellular components of the niche and provide survival signals for incoming and residing HSPC within the bone marrow and are simultaneously osteoblastic progenitor cells that replenish the endosteal niche space. Further relevant cellular components of the marrow environment are regulatory T cells (Tregs), CXCL12-abundant reticular cells (CARs), and endothelial/sinusoidal cells [38]. The effects of cell-cell contact and soluble factors must be discriminated with regard to the potential interference of GvHD effector mechanisms with the bone marrow niche (Fig. 1).

Figure 1.

Flowchart of graft versus host disease-associated myelosuppression. In principle, hematopoiesis can be directly impaired by targeting hematopoietic cells and indirectly impaired by disrupting the cells involved in niche formation. Cell contact-dependent cytotoxicity and soluble factors must be distinguished.

General Considerations

Hematopoietic cells express both classic class I and II HLA proteins and therefore represent potential targets for direct immunocompetent cell responses. Even HSPCs, which reside in an immune-privileged microenvironment, are only partly protected [39, 40]. In mice transplanted with immunocompetent cells (e.g., splenocytes, lymphocytes, or purified subpopulations, especially parent → F1 models), the targeting of host hematopoietic cells via MHC class I and II has been described to result in complete medullary aplasia [39-44]. This GvHD-induced bone marrow failure, which affected all cell lines (including myeloid and lymphoid progenitors as well as stem cells), was identified as the main cause of death in such models. Myelosuppression readily occurred after the infusion of donor lymphocytes [42, 45], and prompt immunosuppressive strategies either prevented or delayed the bone marrow failure [45]. However, delayed immunosuppression could not revert this phenotype, thus indicating irreversible changes [45]. When T-cell-depleted bone marrow was transplanted with syngeneic T cells into MHC-mismatched recipients, medullary aplasia was considerably reduced but not completely interrupted [46] or even absent [47]. This result suggested that donor-derived hematopoiesis might be affected as an innocent bystander under certain circumstances. This collateral damage hypothesis was supported by the observation that the cotransplantation of bone marrow from F1 recipients with medullary aplasia and bone marrow from healthy congenic donors into irradiated F1 recipients induced engraftment failure [41]. In principle, two previously recognized and not mutually exclusive explanations are conceivable, and both would lead to a toxic microenvironment [41], as follows: host-derived nonhematopoietic cells involved in donor hematopoiesis support are attacked (indirect) [48, 49] or donor-derived hematopoiesis is negatively influenced by soluble factors (e.g., cytokines) released during GvHD (direct) [50]. In murine models of GvHD that were induced by the adoptive transfer of immunocompetent cells, both soluble factors, particularly IFNγ, and direct contact-dependent cytotoxicity (e.g., the Fas-Fas ligand [FasL] and perforin-granzyme B pathways) were found to be important mediators of myelosuppression [51]. Medullary aplasia mainly resulted from the targeting of host-derived hematopoietic cells, but host bone marrow stroma cells were also identified as potential targets [41, 42, 48, 52]. In addition to the Fas-FasL pathway, the involvement of other pathways, including tumor necrosis factor (TNF)-related apoptosis-inducing ligand-death receptor (DR) 4/DR5 and TWEAK-DR3, has been reported during the cellular effector phase of GvHD in vivo. Furthermore, inflammatory cytokines such as TNFα and IL-1 play a pivotal role in the amplification of tissue injury during GvHD, but the role of these cytokines in GvHD-associated myelosuppression remains to be elucidated in vivo [51].

The GvHD induced by the adoptive transfer of immunocompetent cells in the aforementioned models mainly recapitulates transfusion-associated GvHD but not GvHD in the context of allogeneic HSCT. In the former case, both hematopoietic cells and niche components are host-derived, while the immunocompetent cells are donor-derived. In the latter case, hematopoietic cells and immunocompetent cells are both donor-derived, while stromal cells remain host-derived [53]. Therefore, direct HLA-mediated targeting of donor hematopoietic cells by donor-derived immunocompetent cells is an unlikely major prerequisite for the development of GvHD as postulated by Billingham (“expression of tissue antigens not present in the transplant donor”) [54]. Consequently, in mice transplanted with both syngeneic immunocompetent cells and stem cells, GvHD-induced myelosuppression is often less impressive and, depending on the mouse strains and the extent of MHC and mHA disparities, other GvHD manifestations might become more prominent [51]. Irrespective of the GvHD model, both the stem cell numbers and functionality have been described as significantly reduced in mice with GvHD (induced by cotransplantation of T-cell-depleted bone marrow and T cells), compared with recipients without GvHD that were only transplanted with T-cell-depleted bone marrow [47, 55, 56]. However, peripheral blood counts, bone marrow cellularity, and myeloid progenitor functionality (as defined by the CFU-GM assay) did not necessarily differ significantly between these two groups [55-57]. Preserved peripheral blood cell counts in mice with GvHD, despite the reduced stem cell capacity, were suggested to be mainly compensated by an expansion of extramedullary hematopoiesis in recipients with GvHD [56]. This finding suggests that the susceptibility to GvHD-associated hematopoietic suppression might vary between different tissues. Nevertheless, pancytopenia occurred in most GvHD models and the bone marrow cellularity dropped significantly [55, 57, 58]. The absolute numbers of lymphoid, myeloid, and erythroid precursors declined as well. This disturbance in maturation was most distinctive in the B-cell lineage, whereas the reductions in nonlymphoid cell populations did not always reach statistical significance [55, 57-59]. In addition, the functionalities of residual B-cell precursors and mature B lymphocytes were reduced [57]. In summary, GvHD affected the overall process of hematopoiesis but particularly focused on the B-cell lineage, leading to a relative shift toward myelopoiesis. This impairment of B lymphopoiesis in the bone marrow was shown to rely predominantly on MHC class II-mediated donor CD4+ T-cell activity in mice with mutant MHC class I and class II alleles that received purified CD4+ and CD8+ T cells, respectively, and in GvHD mice treated with anti-CD4 and anti-CD8 monoclonal antibodies, respectively [55].

Cell-Contact-Dependent Myelosuppression

The Fas-FasL pathway has been suggested to play a decisive role in GvHD-associated myelosuppression and lymphoid hypoplasia. In MHC-matched murine GvHD models, the cotransplantation of FasL-defective T cells or treatment with monoclonal antibodies that block the Fas-FasL interaction was found to reduce lymphopenia. The cotransplantation of FasL-defective T cells also ameliorated myelosuppression; however, this phenomenon was not observed when the Fas-FasL interaction was blocked, likely because of the pharmacokinetic properties of the monoclonal antibody [55, 59, 60]. In contrast, the transplantation of perforin-deficient T cells did not reduce the severity of splenic lymphoid hypoplasia [59]. The transplantation of FasL defective T cells also gave rise to incomplete donor chimerism, indicating that persistent host stem cell destruction depends on the Fas-FasL pathway [59]. Fas-induced apoptosis was suggested to be responsible for lymphoid hypoplasia and myelosuppression, and the targeting of donor-derived hematopoietic cells in the bone marrow was thought to be mediated either by inflammatory cytokine-induce (e.g., TNFα and IFNγ) Fas overexpression on HSPCs or by elevated FasL expression on alloactivated host-derived T cells (bystander killing) [59]. In accordance with this suggestion, the favorable effects of inhibiting Fas-induced apoptosis on lymphoid hypoplasia were even more pronounced in IFNγ receptor-deficient (IFNγ-RKO) recipients [60]. However, mice transplanted with Fas-defective bone marrow (resistant to Fas-mediated cytotoxicity) and wild-type T cells still suffered from B lymphopoiesis suppression [55] and exhibited only gradually ameliorated histological bone marrow GvHD scoring [58], thus arguing against predominant Fas-mediated bystander killing, as described previously. This suggestion was confirmed in retransplantation assays. Mice that were transplanted with bone marrow from GvHD-affected donors subsequently showed no defects in B lymphopoiesis. In contrast, GvHD mice that were retransplanted with healthy bone marrow showed a persistent disturbance in B lymphopoiesis. Comparable results were obtained when mice transplanted with Fas-deficient bone marrow showed no significant improvements in the GvHD-associated histological changes. These results confirmed the importance of the Fas-FasL pathway in GvHD-associated myelosuppression but also suggested that the hematopoietic niche, rather than HSCs, was targeted by GvHD. Indeed, in mice with chronic GvHD, the absolute number of osteoblasts was drastically reduced and the ability of the remaining osteoblasts to form bone was diminished. In contrast, the absolute numbers of endothelial cells and CXCL12-abundant reticular cells were not reduced, but endothelial permeability had increased. Although the functionality of these cells with regard to HSC support was not determined, the osteoblastic niche was suggested to represent the main target of GvHD-associated myelosuppression. As determined by immunofluorescent staining, osteoblasts did not express MHC class II, suggesting a MHC-independent pathway for the induction of cytotoxicity [55]. This result contradicts the aforementioned observation that the disruption of B lymphopoiesis largely depends on MHC class II. This difference may be explained methodically, but also gives rise to speculations about other involved cell types. MSCs possess multilineage potential and can differentiate along the osteogenic, chondrogenic, and adipogenic lineages [61] and are therefore involved in the formation of the HSPC niche [37]. To mimic an inflammatory milieu reminiscent of the alloreactivity that occurs during mismatched allogeneic HSCT, MSCs were incubated in mixed lymphocyte reaction (MLR) supernatants that contained high concentrations of TNFα and IFNγ. The proliferative activities of the MLR-treated MSCs increased, and differentiation shifted toward the osteogenic lineage. The ability of the MLR-treated MSCs to support HSCs in vitro was reduced. Interestingly, MHC class II expression on MLR-treated MSCs was significantly upregulated [62], and this process was shown to be IFNγ dependent [62-64]. These findings suggest that the inflammatory milieu associated with GvHD can impair hematopoiesis by modifying the function of MSCs. This process might be still further enhanced by an increased susceptibility to allogeneic targeting consequent to the increased MHC class II expression [62]. Consistent with the hypothesis of niche-GvHD, clear reductions in the bone marrow stromal cell and osteoblast populations have been described in patients with GvHD [49, 65, 66].

Myelosuppression Mediated by Soluble Factors

The role of IFNγ in GvHD-associated myelosuppression was further explored in IFN-RKO recipients that were transplanted with bone marrow from wild-type mice. Consequently, the host-derived cells were unresponsive to IFNγ, whereas the donor cells were exposed to exceptionally high levels of IFNγ. GvHD symptoms, which were clearly associated with host-derived tissue damage, increased in some organs, including the hematolymphoid system. In contrast to the Fas-induced apoptosis observed in lymphoid tissues, donor-derived hematopoiesis suppression resulted from reduced proliferation rather than apoptosis induction. Nevertheless, this hematolymphoid reconstitution failure could be corrected by neutralizing IFNγ [60]. In addition, the GvHD-associated histological changes in the bone marrow of recipient mice were significantly ameliorated upon the transplantation of IFNγ-RKO bone marrow, thus clearly demonstrating the IFNγ dependence of this process [58]. One possible explanation is a direct inhibitory effect, given that HSC exposed to IFNγ in vitro exhibits a compromised self-renewal capacity [67]. Furthermore, the deleterious effect of IFNγ might be readily explained by the upregulation of MHC expression on the host-derived cells and the enhanced propagation of donor-derived cytotoxic lymphocytes. However, IFNγ primarily seems to mediate protective effects in GvHD. Besides its divergent effects on target and effector cells in various tissues [68], the opposing effects of IFNγ, which depend on the intensity of the conditioning regimen, might explain these ostensibly conflicting results [69].

Although, TNFα is involved in multiple stages of the GvHD process [3, 51] and has been shown to negatively regulate HSCs both in vitro and in vivo [70-72], the importance of TNFα in GvHD-associated myelosuppression has not yet been validated in detail. Because TNFα also alters the functionality of MSCs [62], it is conceivable that TNFα may additionally harm HSCs in an indirect fashion.

Pathophysiological Conclusions and Questions

In conclusion, the GvHD-associated suppression of donor-derived hematopoiesis in allogeneic HSCT is induced by the toxic microenvironment (Figs. 1 and 2). IFNγ and TNFα disrupt donor hematopoiesis in both direct and indirect manners. These cytokines impair the functionalities of donor-derived hematopoietic cells and host-derived MSCs. In addition, IFNγ increases MHC class II expression on MSCs, thus rendering them more susceptible to allogeneic targeting. Host-derived osteoblasts are targeted via the Fas-FasL pathway (cell-contact-dependent cytotoxicity), resulting in the indirect impairment of donor-derived hematopoiesis. However, important questions remain to be elucidated. For example, are recipient osteoblasts recognized independently of MHC? Do other cell-contact cytotoxicity pathways participate in the targeting of donor-derived stromal cells? Are other niche components involved? Do other soluble factors contribute to impaired donor hematopoiesis? Is there a role for cell-contact cytotoxicity in the targeting of donor hematopoiesis?

Figure 2.

Model of graft versus host disease-associated myelosuppression. Donor-derived alloreactive T cells are activated by host-derived APCs. TNFα, which is also released during tissue injury, and IFN-γ are secreted by activated effector cells. TNFα and IFNγ impair the functioning of host-derived stromal cells and donor-derived hematopoietic cells, including HSCs (cytokine milieu). IFNγ induces the expression of MHC class II on host-derived MSCs, thus increasing the susceptibility of these cells to alloreactive cell-contact-dependent killing. Host-derived osteoblasts, which are involved in niche formation, are targeted via the FasL pathway in a MHC-independent manner (alloreactive killing). Perforin-granzyme B pathway involvement in cell contact-dependent cytotoxicity within the niche is controversial. The cell contact-dependent killing of donor hematopoietic cells by cytokine-induced overexpressed FasL on donor-derived effector cells and Fas on donor-derived hematopoietic cells has been suggested (bystander killing). The disrupted niche and soluble factors form a toxic microenvironment for donor-derived hematopoiesis. Abbreviations: APCs, antigen-presenting cells; FasL, Fas-Fas ligand; HSCs, hematopoietic stem cells; IFN, interferon; MHC, major histocompatibility complex; MSCs, mesenchymal stromal cells; TNFα, tumor necrosis factor-alpha.

Clinical Relevance and Therapeutic Perspectives

The perturbance of B- and T-cell development and function has been described in both acute and chronic GvHD. The consequent immunosuppression and susceptibility to infection (aggravated by the use of concomitant immunosuppressive therapy) have been identified as important risk factors. GvHD-associated myelosuppression substantially contributes to this lymphoid impairment in addition to disturbances in the secondary lymphoid organs. Thrombocytopenia, which acts as an independent predictor of worse outcomes, further emphasizes the importance of GvHD-associated myelosuppression [19]. However, clinical studies have rarely focused on the reversibility of myelosuppression during GvHD therapy and to date, no clinical studies on the treatment of bone marrow GvHD have been published. Currently, corticosteroids comprise the only accepted general first-line therapy for acute and chronic GvHD, and there is no established second-line therapy [3].

Most therapeutic approaches have focused on effector cells. Unfortunately, these attempts have not consistently improved the outcomes of patients with steroid-refractory GvHD [31] (Table 1 and Fig. 3). In addition to drug-specific side effects, the increased risks of graft failure and infections, which are amplified by effector phase inhibition, contribute to poor outcomes in steroid-refractory GvHD cases.

Table 1. Potential therapeutic approaches in bone marrow graft versus host disease
Targeting the effector phase of GvHD
  1. Classification according to the involved cell compartments.

  2. Abbreviations: ATG, anti-thymocyte globulin; CsA, cyclosporin A; ECP, extracorporeal photopheresis; ESA, erythropoiesis-stimulating agents; G-CSF, granulocyte colony-stimulating factor; GvHD, graft versus host disease; HSPCs, hematopoietic stem and progenitor cells; IL-2, interleukin 2; IMPD, inosine-5′-monophosphate dehydrogenase; MSCs, mesenchymal stromal cells; mTOR, mammalian target of rapamycin; n.a., not available; NKT, natural killer T cells; TNFα, tumor necrosis factor-alpha; Tregs, regulatory T cells; TPO, thrombopoietin.

Suppression of effector cellsLymphodepletive chemotherapyPentostatin, methotrexate
ImmunotherapyATG, alemtuzumab, denileukin-diftitox
Inhibition of function of effector cellsCalcineurin-inhibitorsCsA, tacrolimus
mTOR inhibitorsSirolimus, everolimus
IMPDH inhibitorsMycophenolat mofetil
ImmunomodulationECP, Tregs, MSCs, NKT
Neutralization of soluble factorsIL-2 interferenceDaclizumab, denileukin diftitox
TNFα interferenceInfliximab, etanercept
Targeting HSCPs
MyelopoiesisG-CSFLenograstim, filgrastim
ErythropoiesisESAEpoetin alpha, epoetin beta, darbepoetin
ThrombopoiesisTPORomiplostin, eltrombopag
Targeting the niche
 Protection of niche functionTregs
 Support of niche functionMSCs
Figure 3.

Potential therapeutic approaches to graft versus host disease (GvHD)-associated myelosuppression. Drugs that stimulate different hematopoietic lineages (e.g., TPO-mimetics, ESA, and G-CSF) might be useful for the supportive care of patients with GvHD-associated cytopenia in order to improve the quality of life and reduce transfusion requirements. To date, most noncellular immunosuppressive strategies have focused on targeting and inhibiting effector cells. Regulatory T cells (Tregs) efficiently suppress effector cell function and have been described to provide immune privilege to the hematopoietic stem and progenitor cell niche. However, clinical experiences with the adoptive transfer of Tregs are very limited. MSCs could theoretically provide several clues for the treatment of GvHD-associated myelosuppression (e.g., effector cell inhibition, tissue repair, and niche function restoration) but to date, the successful translation of MSC-based therapy to clinical medicine has failed (controversial results). Adequate homing of the MSCs to the bone marrow after adoptive transfer has not yet been observed. Abbreviations: ESA, erythropoiesis-stimulating agent; G-CSF, granulocyte colony-stimulating factor; HSCs, hematopoietic stem cells; IFN, interferon; MHC, major histocompatibility complex; MSCs, mesenchymal stromal cells; TNFα, tumor necrosis factor-alpha; TPO, thrombopoietin.

Drugs that have been approved for stimulating individual hematopoietic lineages except for B and T cells are commercially available and are routinely used beyond HSCT. Even in the context of allogeneic HSCT, the serum levels of erythropoietin (EPO), thrombopoietin (TPO), and granulocyte colony-stimulating factor (G-CSF) have been demonstrated to correlate inversely with the red blood cell, platelet, and neutrophil counts, respectively, thus reflecting normal physiological behavior [35, 73-78]. Only EPO levels have been described as inadequate for the anemia observed after MAC [73, 77]. No clinical trials of erythropoiesis-stimulating agents (ESAs), TPO-mimetics, or G-CSF for GvHD-associated anemia, thrombopenia, or neutropenia, respectively, have been published. However, G-CSF is routinely recommended for GvHD-associated neutropenia [3] because it could theoretically improve survival by reducing the risk of infectious complications. Regarding ESAs, there have been conflicting historical reports on the usefulness of these agents in individuals with GvHD [35, 79]. Interestingly, the TPO serum levels were observed to be reduced in patients with acute GvHD [75], and a case series of patients with SFPR (four of seven in the context of GvHD) reported successful platelet recovery (>50,000/µl) in all patients treated with romiplostim [80]. Nevertheless, it is doubtful whether TPO-mimetic therapy could confer improved survival upon patients with GvHD, as thrombopenia is not associated with an increased risk of death due to bleeding but rather represents a general surrogate endpoint of myelosuppression and GvHD. However, TPO-mimetics might improve the quality of life and reduce the need for transfusion in such cases.

Tregs and MSCs display immunomodulatory properties both in vitro and in vivo. The exact mechanisms remain to be revealed in both cell types, but their functions apparently depend on both soluble factors and cell-cell contact. The ability of these cell types to suppress effector T-cell proliferation and function suggests that they might be promising cellular therapeutics for immune-mediated diseases such as GvHD. To date, only limited clinical data are available regarding patients with acute and chronic GvHD [81, 82]. In particular, MSC-mediated GvHD treatment and prophylaxis yielded conflicting results, possibly due to differences in the MSC sources, isolation techniques, ex vivo expansion methods, storage conditions, dosages, and administration schedules [81, 83]. Nevertheless, Tregs and MSCs might be of special interest for the treatment of bone marrow GvHD, as both have described to be involved in niche formation and HSC functional maintenance, respectively [37, 39]. Upon transplantation into unconditioned mice, allogeneic HSCs persisted in the HSPC-niche without immunosuppression and retained their functionality for at least 30 days, whereas more committed cells were rapidly rejected. The failure to reject allogeneic HSCs was shown to depend on the functions of host-derived Tregs, which accumulated in proximity to the HSPC niche [39]. De novo generated donor-derived Tregs were also reported to accumulate in the bone marrow of allogeneic stem cell recipients [84], but the homing of adoptively transferred allogeneic Tregs to the bone marrow has not yet been reported. MSCs theoretically represent an optimal therapeutic population for bone marrow GvHD, as these cells exhibit immunosuppressive, regenerative, and HSC-supportive properties. However, as the biology and function of MSCs remain largely unclear, the adequate translation of these cells into clinical applications is pending. To date, bone marrow MSCs were shown to remain of host-origin after HSCT, and the homing and engraftment of adoptively transferred MSCs to the bone marrow have not been observed at reasonable levels [83]. Finally, caution must be suggested, as the protection of niche function is not free of potential risks; for example, leukemic stem cells have been shown to hijack the bone marrow niche [85, 86]. Therefore, niche protection might permit leukemia relapse, particularly when concomitant immunosuppressive therapy acts to suppress the effects of GvT.


M.B. is supported by the Deutsche Forschungsgemeinschaft (SFB 655, project B2).

Author Contributions

M.v.B.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; M.B.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.