Recently, CD4+ CD25+ FOXP3+ Tregs have gained attention for their immunosuppressive capabilities in the AML tumor microenvironment. Accumulating evidence has shown that proportions of Tregs are increased in the peripheral blood and bone marrow of AML patients and are associated with poorer prognosis (41, 60–63). In addition, current research suggests that Tregs promote leukemia cell survival by suppressing both expansion and effector functions of cytotoxic T lymphocytes, as well as directly promoting the stemness and survival of AML blasts through the IL10/PI3K/AKT signaling pathway (64–66).

Tregs are a heterogenous population and can be both naturally derived in the thymus (nTregs), which is believed to make up approximately 80% of Treg repertoire, or induced by conversion of conventional T cells existing in the periphery (iTregs). Whether Tregs within the tumor microenvironment are natural vs induced remains a subject of debate, as they are difficult to distinguish in vivo (67). However, several biomarkers such as Helios, NRP1, and FOXP3 methylation status have been reliably used to identify Tregs of thymic origin (68–70). In this regard, a previous study showed that expanded Tregs from AML patients treated with histamine dihydrochloride (HDC) and low dose IL-2 for relapse prevention predominantly resembled nTregs, based on Helios expression and TSDR methylation status in the FOXP3 gene locus (71). These findings corroborate a previous study showing that a majority of tumor-infiltrating Tregs are thymically derived and reactive to Aire-dependent self-antigens associated with the organ of cancer origin (72). Together, these studies have important clinical implications because thymic involution from aging or chemotherapy has been shown to enhance the generation of nTregs (73).

As we have previously discussed, thymus involution has been known to result in limited TCR diversity. Several groups have similarly reported restricted TCR repertoire diversity and skewed oligoclonal αβ⁺ and γδ⁺ T- cell pools in AML (74–77). Notably, recent studies have shown restricted expression and oligoclonal expansion of certain Vβ subfamily T cells, such as PD-1⁺ Vβ T cells in patients with AML and CML (76, 77). The oligoclonal expansion of certain TCR γδ T cells from patients with AML have likewise been reported (78). Although γδ T cells are less studied and constitute a smaller percentage of the human T cell population, there is growing evidence to support its role in antitumor immunity (79). In fact, Vγ9Vδ2 T cells have been shown to specifically recognize and kill AML blasts in a TCR dependent manner (80). A study comparing TCR repertoires between pediatric and adult AML showed that the fraction of γδ T cells increases with age, which is the inverse of what is seen in healthy patients (74). Taken together, it is likely that certain leukemic-specific T cell clones in the TME may be preferentially expanded by AML-associated antigens in the periphery and/or are being restricted during development in the thymus. In support of the latter, a recent study found that effector functions of γδ T cell subsets within the TME was dictated by metabolic programing during early thymic development (81).

Age-related thymic atrophy has been correlated to the development of systemic inflammation with advanced age, termed inflammaging (45). Recent studies also suggest that mTEC dysfunction with age reduces the capacity to eliminate self-reactive thymocytes through impaired negative selection (82, 83). This process has been proposed to potentiate the increasing self-reactivity and auto-inflammation observed with aging (45). It is known that chronic inflammation can aid in tumor progression, metastasis and drug resistance in tumor cells (84), and pro-inflammatory mediators also promote the disease progression and relapse of AML (16, 85). Accordingly, it is thus possible that the disruption in T cell homeostasis derived from thymic atrophy might play a role in AML pathogenesis.

Previous studies have shown that CD8+ T cells are functionally impaired in AML including expressing aberrant phenotypes, having increased expression of exhaustion/senescence molecules (CD57, TIGIT, TIM-3 and PD-1), and forming defective immune synapses with blasts (11, 12, 86–90). While thymic function gradually decreases with age, continued thymopoiesis in residual thymus tissue has been shown to be essential for the generation of new CD8+ T cells to replenish attrition from exhausted T cells during chronic antigen exposure. This has been demonstrated in patients with chronic polyomavirus infection who depend on the reactivation of thymic function and de novo T cell generation to sustain antigen-specific CD8+ T cell immunity (91).

The thymus is exquisitely sensitive to cytoreductive therapies that are traditionally used as the standard of care in AML, such as chemotherapies or radiation therapies. While all hematopoietic cell types are affected following chemotherapy, T cells appear to be most profoundly impacted, likely in part due to depletion of thymocytes and TECS within the thymic stromal compartment (109, 110). The recovery of the depleted T cell pool after cytoreductive therapy may take many years or be permanent (111, 112). Age-related thymic involution has also been linked to impaired immune reconstitution following chemotherapy, which may partially explain treatment failure in the elderly (113–115). Considering the impact of cytoreductive therapies on thymic function and de novo T cell output, adoptive T-cell based immunotherapies utilizing bioengineered allogenic or autologous T cells, such as CAR T cell therapy or tumor-infiltrating lymphocytes (TILs) may be a promising therapeutic approach, either as a supplement to first-line treatments or as a treatment for refractory cases (14, 116).

Relapsed and refractory disease presents a significant challenge and is primarily responsible for the poor prognosis in AML. The emergence of remnant populations of leukemic cells after exposure to chemotherapy, termed measurable residual disease (MRD), implies that AML cells may develop drug-resistance or mechanisms to evade exposure to therapy. Several premetastatic reservoirs, such as the bone marrow and the perivascular space of blood vessels, have been identified as potential reservoirs that permit primary tumor cells to hide following adjuvant chemotherapy (117–119). When damaged by chemotherapy, the atrophied and inflamed thymus has likewise been found to create a chemoprotective microenvironment for both solid and hematologic tumor cells (120–122). Notably, the release of the proinflammatory cytokines TIMP1 and IL-6 from TECs in response to cytotoxic chemotherapy has been shown to induce TEC senescence and promote lymphoma cell survival (121). Exposure to chemotherapy also induces thymic residing tumor cells to acquire an antiapoptotic chemo-resistant phenotype within the inflammatory thymic microenvironment (120). Finally, the presence of tumor cells within the thymus may also interfere with de novo T cell differentiation and lead to tumor-specific immune tolerance. This hypothesis is supported by several studies that have shown that infectious pathogens that invade the thymus lead to microbe-specific T cell tolerance (123–127). Together, these findings suggest that the atrophied thymic microenvironment may be chemoprotective to AML blasts, harbor MRD following adjuvant therapy, and result in tumor-specific T cell tolerance that leads to an eventual tumor relapse.

FMS-like tyrosine kinase 3 (FLT3) is widely expressed by immature hematopoietic progenitors and over-expressed in a majority of AML blasts, often in the presence of activating mutations by tandem duplications (FLT3-ITD) and/or point mutations involving the tyrosine kinase domain (TKD) (128). The presence of FLT3-ITD is widely accepted to portend a poor prognosis in AML due to chemoresistance and the high rate of relapse. Recent data has suggested that FLT3 and its respective ligand (FLT3L) may be an important point of regulation for thymic function. FLT3L knockout mice were notably observed to have reduced immature thymocytes and Lin^-SCA1⁺KIT⁺ (LSK) lymphoid-primed multipotent marrow progenitors (129, 130). More importantly, administration of FLT3 ligand was found to enhance the export and survival of LSK cells, early thymic progenitors after transplantation, and androgens via the downregulation of CXCR4 (131). Potentially, these findings have important clinical implications because the use of FLT3 inhibitors after HSCT may impair thymic recovery in AML.

Allo-HSCT is currently the only curative treatment option for AML; though, its success both in regards to its ability to fight opportunistic pathogens and eliminate leukemia depends on optimal recovery of adaptive immunity after cytoreductive conditioning (132). Thymic-independent proliferation of donor-derived T-cells can initially restore peripheral T-cell numbers within the first year following Allo-HSCT. However, because these T-cells are derived from a limited number of donor precursors, their TCR repertoire is of limited diversity (133–136). This has important clinical implications, as lower TCR diversity has been associated with increased risk for disease relapse (137). Therefore, complete T-cell reconstitution with a diverse TCR repertoire may depend on the recovery of de novo T-cell production in the recipient thymus (132, 136, 138). A recent study has importantly demonstrated that some donor-derived CD8 T cells specific for hematopoietic cell-restricted antigen can escape deletion in the thymus and contribute to GVL effects in the periphery (139). Considering that thymic function is also highly sensitive to pre-transplant conditioning regimens, the above findings provide a rationale to developing strategies to boost thymic recovery to improve the outcome of Allo-HSCT (140).

GVHD is a common but significant complication of Allo-HSCT and has additionally been shown to interfere with thymic-dependent T cell development by reducing TSP as well as altering thymic cellularity and architecture (138). This is in part mediated by a glucocorticoid-independent mechanism of DP thymocyte apoptosis, as well as the destruction of mTECs via donor alloreactive T cells by expression IFN-gamma, and the cognate proteins FasL and TNF-related apoptosis-inducing ligand (TRIAL) (138, 141–147). The capacity to affect thymic output after HSCT has been demonstrated by the observation that acute GVHD (aGvHD) is associated with a significant reduction in sjTREC counts (148–150). In chronic GvHD (cGvHD), research has demonstrated that sjTREC levels remain reduced even after recovery and long-term follow-up. However, this effect was transient in young patients <25 years old, highlighting the age-dependent regenerative capacity of the thymus (150).

Therapeutic cancer vaccination to induce remission or prevent relapse has been extensively explored in preclinical and clinical trials over the last decade (151). However, the efficacy of tumor vaccination strategies may be compromised by several tumor extrinsic factors secondary to thymic atrophy, including Treg driven immune suppression, T cell exhaustion, and reduced TCR repertoire. Though the role of thymic atrophy on anti-tumor vaccines has yet to be studied directly, age-related thymic involution has been cited as one cause for the diminished response to traditional vaccination. This has been suggested based on the observation that only an estimated 30-40% of elderly patients are able to mount an adequate immune response to the influenza vaccine (152).

The success of immunotherapy in AML is contingent on the presence of functional T cells capable of recognizing tumor specific antigens. Additionally, the degree of neoantigen heterogeneity, tumor mutational burden, and TCR repertoire diversity has been demonstrated to increase the likelihood of tumor antigen recognition and improve overall survival (153–155). Considering the role of the thymus in maintaining homeostatic T-cell immunity, it is likely that a decline in its function could compromise the efficacy of T-cell directed immunotherapies in AML.

The use of immune checkpoint inhibitors (ICIs) in AML, such as anti-CTLA-4 antibody ipilimumab, and the anti-PD-1 antibody nivolumab have demonstrated limited response compared to its impressive efficacy in solid tumors (156–158). Several strategies have been employed to predict response to ICI in solid tumors, including TCR repertoire profiling by high-throughput sequencing before and after ICI treatment (159). Previous studies in melanoma, gastrointestinal cancers, and relapsed/refractory classical Hodgkin lymphoma have overall shown that a broader TCR profile is associated with superior outcomes in patients receiving ICIs (160–164). Consistently, a recent paired analysis of single cell RNA and TCR repertoire profiling in relapsed/refractory AML demonstrated that TCR repertoires primarily from CD8+ cells expand in patients who responded to PD-1 treatment, but contract in those who were treatment resistant (165). Perhaps assessment of TCR repertoire and even the functional status of the thymus can be used to risk stratify and monitor patients receiving ICI therapy and thereby determine optimal personalized treatment strategies for AML patients.

While the application of genetically engineered CAR-T-cell therapies have been successful for lymphoid malignancies, its use for the treatment of heterogeneous AML faces several unmet challenges (166). This includes the lack of AML-specific cell-surface antigen that minimizes off target toxicity and the suppression of T cell activity and proliferation by AML blasts (167). As previously discussed, the increased Treg frequency following thymic atrophy creates a hostile tumor microenvironment for CAR T cells who are already susceptible to exhaustion via chronic tumor antigen exposure. There is some research to suggest that mature peripheral T cells can re-enter the thymus, localize to the medulla, and alter the stromal microenvironment (168, 169), which was attributed to the onset of thymic involution (170). Remarkably, adoptively transferred syngeneic antigen-specific T cells in lymphopenic mice has been shown to enter the thymus and eliminate thymic dendritic cells (DCs) and mTECs presenting cognate antigen (171). These findings have important clinical implications, as CAR T cells may eradicate tumor antigen expressing APCs that mediate negative selection, thereby enhancing the patient’s endogenous antitumor repertoire. However, given that most AML antigens may also be expressed by normal HSPCs and healthy tissue, caution must be used, as this phenomenon may also augment the risk for autoreactivity via on-target, off-tumor toxicity (116).

As previously discussed, the thymic stromal compartment is crucial for intrathymic T cell development. The expression of key transcription factor, Forkhead box N1 (FOXN1) is important for TEC differentiation, thymic organogenesis during embryonic development, antigen processing, thymocyte selection, and has been implicated in regulating age-related thymic involution (257–260). While downregulation of foxn1 gene with age results in functional decline in the TEC compartment, its overexpression has been shown to delay thymic degeneration (257, 261). A study by Bredenkamp and colleagues showed that increased FOXN1 expression is sufficient to drive regeneration of the aged thymus both in regards to its architecture, gene expression and functionality (192). More recently, Oh et al. demonstrated embryonic fibroblasts reprogrammed to overexpress FOXN1 could be engrafted to rejuvenate thymic function in mice (193). While there are currently no ongoing clinical trials, developing therapies which target FOXN1 expression may provide one strategy for regenerating an aged thymus further adversely affected by chemotherapeutic regimens used to treat AML.

Interleukin-7 has come into focus as an important non-redundant regulator of lymphopoiesis and mature T cell homeostasis through both thymic dependent and independent mechanisms (10, 262). In the thymus, IL-7 promotes the survival of DN thymocytes, TCR rearrangement, and lineage differentiation during positive selection (263). Furthermore, IL-7 expression appears to decline with age, correlate with thymic atrophy, and importantly appears to be down-regulated in the peripheral blood of AML patients (264–266). The use of exogenous IL-7 for immune reconstitution has been extensively investigated in several preclinical and clinical studies (262). Administration of IL-7 in mice after T-cell depleted allo-HSCT was found to significantly expand donor-derived thymocytes and peripheral T cells, but remarkably had no effect on alloreactive T cells and the development of graft-versus-host disease (206, 207). Several clinical trials have also evaluated the use of glycosylated recombinant human IL-7 (CYT107) in HIV-1 infected patients and found that IL-7 treatment was not only safe and well tolerated, but could also enhance thymopoeisis as demonstrated by an increased number of recent thymic emigrants (RTEs), increased TREC ratio, and increased TCR repertoire diversity (208–210). Notably, these findings were addressed in a phase 1 clinical trial in which CYT107 was used in recipients of T-cell depleted allo-HSCTs; unfortunately, changes in RTE or TREC levels were only appreciated in a small subset of patients (211).

Keratinocyte growth factor (KGF) is a potent mitogen expressed by thymic mesenchymal and stromal cells that mediates TEC proliferation and survival through activation of the PI3k-AKT-nuclear factor-kB and p53 pathways (176, 177). Several studies have demonstrated that treatment with exogenous KGR could avert GVHD-related injury, enhance thymopoiesis in HSCT recipients and protect TECs during irradiation-induced injury (176, 267, 268). Furthermore, the use of KGF could reverse age-related thymic involution and restore thymopoiesis in aged mice for up to 2 months after treatment (269). The use of recombinant KGF (palifermin) has since been approved by the US Food and Drug administration for the treatment of oral mucositis in patients receiving intensive chemotherapy (270). Currently, there are multiple clinical trials underway to further investigate the use of palifermin for T cell reconstitution (NCT01233921, NCT03042585, NT02356159 and NCT00593554).

The observation that thymic function rapidly declines after puberty and that castration rejuvenates thymic function has suggested that sex hormones, particularly androgens, play a role in thymic involution (48, 271–273). Consistent with these findings, several studies have demonstrated that sex steroid inhibition (SSI) promotes thymopoeisis by increasing TEC expression of CCL25, Dll4, and Notch signaling pathways (223, 224). The regenerative impact of SSI has further been found to enhance immune function in immunocompromised patients or those undergoing auto-HSCT (274, 275). To date, several drugs have been developed to transiently inhibit sex steroids, such as luteinizing hormone-releasing hormone (LHRH) antagonists, for use in precocious puberty, prostate cancer, breast cancer, and endometriosis. More recently, two clinical trials (NCT01746849 and NCT01338987) are underway to evaluate the effects of SSI on immune reconstitution following allo-HSCT. A pilot study using LHRH agonist (goserelin) administration 3 weeks prior to HSCT has already demonstrated significant increases in naive CD4+ T cells, TRECS, and recovery of TCR repertoire diversity (225).

The expression of growth hormone (GH) progressively declines after the third decade of life and has been linked to both hematopoietic and thymic function (194, 276, 277). Given its immunomodulatory effect in humans, several preclinical and clinical trials have investigated the use of exogenous GH to boost immune function and demonstrated that it can reverse thymic atrophy and improve TCR diversity (194–196). Notably, two prospective randomized trials reported that daily recombinant GH injections for 6-10 months could enhance thymic output and TREC levels in HIV-1 infected patients (195, 278). Another trial also remarkably showed that GH treatment reduced PD-1⁺ CD8 T cells, suggesting that this could also be used to reverse CD8 T cell exhaustion (197). While GH therapy has been approved for use in pediatric patients who develop post-radiation growth disorders after HSCT, GH therapy has yet to be implemented for immune reconstitution (279, 280). Before it may be used for thymic reconstitution, several concerns must be addressed. Firstly, it appears the effect on thymic output appear to only be transient, as discontinuation of GH results in recurrence of thymic atrophy (196). Secondly, GH is associated with many undesirable side effects, including increased risk for cardiovascular disease, stroke, and diabetes. Importantly, GH and growth hormone-releasing hormone (GHRH) have recently been implicated in AML proliferation (281, 282). These side effects reduce the enthusiasm for the use of GH or GHRH in treating AML.

The expression of Notch ligands Delta-like-1 and 4 (DLL4 and DLL1) by cTECs is essential for supporting T-lineage commitment (283). In the absence of notch signaling, thymopoiesis is arrested early in T cell development during double negative thymocyte differentiation (284). Conversely, its constitutive activation in HSPC simultaneously results in inhibition of B cell development and promotion of T cell development toward double positive status in the thymus (285). Furthermore, a recent study by Tikhonova et al. investigating the transcriptional changes of the BM microenvironment in response to chemotherapy demonstrated that DLL4 and DLL1 is notably downregulated by the vascular endothelium in response to stress (286).

The potential for Notch signaling for T cell reconstitution has been explored in several studies. For instance, the treatment of lin^- Sca-1⁺ C-kit^- (LSK) hematopoietic progenitors with DLL1 ex vivo has been shown to accelerate thymus engraftment and T cell reconstitution after HSCT (204). More recently, it has been shown that activation of a BM-specific Notch/IL-21 signaling axis could lead to ex vivo expansion of T cell progenitors (205). It is important to note that Notch has also been cited as a tumor suppressor in myeloid malignancies and that reduced Notch signaling may play a role in skewing HSPC to premature myeloid lineages that could progress to leukemic myeloid production (287–289). Together, these findings provide a rationale to develop strategies to improve for T cell regeneration that utilize Notch and its downstream signaling pathways for use in treatment of AML.

Bone morphogenic protein-4 (BMP4) is a member of the TGF-β superfamily, known for its role in regulating embryonic development (290). However, new research has highlighted its function in thymic tissue regeneration after injury (200, 201). A recent study by Wertheimer et al. showed that TECs increased expression of BMP4 after thymic injury, resulting in increased expression of Foxn1 and its downstream target, Notch ligand Dll4 (200). More remarkably, they found that inhibition of BMP4 resulted in impaired thymic regeneration, and its exogenous administration could rescue its ability to repair. While not yet in clinical trials, these studies highlight BMP4 as a novel target for improving immunologic recovery.

Adoptive cell-based therapies and de novo thymus synthesis have also been investigated for the purpose of enhancing thymic function. Considering that T cell reconstitution is delayed after HSCT due to the limited availability of hematopoetic progenitors, several groups have investigated the potential of supplying donor precursor T cells to boost thymogenesis at the time of transplant (247–249). By expanding hematopoietic precursor cells ex vivo using Notch-1 stimulation, researchers have demonstrated significant increases in both thymic cellularity and peripheral T cell reconstitution (250–252). Alternatively, others have attempted to generate new thymus epithelial tissue through the isolation and expansion of thymic epithelial progenitor cells (TEPC) from fetal thymi (241, 242).

Here, we have extensively reviewed the regenerative factors that are currently being studied for repair of the dysregulated immune system, some of which are now being investigated in clinical trials ( Table 1 ). We acknowledge the challenge in developing a successful and durable effective immunotherapy for heterogeneous AML in elderly patients, based on the rationale that their declined immune function leads to further disease progression (291). Thus, novel approaches will also be crucial. It has long been recognized from parabiotic mouse studies that there are circulating factors in young mice, which have not yet been identified but which could improve the metabolic and regenerative status of older mice (292). It will then be important to determine if the immune system in the aging mouse also improves with parabiosis. If so, parabiosis could be adapted to studies on the dysregulated immune system in a humanized AML mouse. A serial chronological study of the immune system utilizing the latest genomic techniques, such as single-cell RNA-seq and epigenetic approaches, could disclose key pathways involved in the deterioration of immune regulation with aging and AML. Such mechanistic studies and translational approaches could be applied to both the bone marrow niche and the thymic microenvironment. Searches for important molecular and genetic interactions could be sought as well.