Acute myeloid leukemia (AML) is a hematological malignancy which interferes with the normal haematopoietic production of blood cells. It is characterized by the accumulation and expansion of immature myeloid cells (blasts) within the bone marrow that proliferate rapidly and inhibit differentiation of the normal progenitor cells (1). AML is primarily a disease of adulthood and comprises ~80% of hematological malignancies observed in patients over 60 (2, 3). AML is typically treated by a standard induction chemotherapy of a 7 + 3 regimen of cytarabine with an anthracycline (daunorubicin or idarubicin), genetic risk profiles then determine follow up treatment of consolidation chemotherapy or allogeneic hematopoietic stem cell transplantation (4). The efficacy of standard induction chemotherapy is limited and is followed by AML relapse and mortality in a majority of patients over 60 (4, 5).

The “two-hit” model of AML leukaemogenesis proposes that two mutation classes drive the disease, those that augment proliferation and cell survival (e.g., FLT3) and those which disrupt haemopoietic differentiation and apoptosis (e.g., CBFβ-MYH11) (6). Recent development of therapies targeting such mutations including the FLT3 inhibitors, lestaurtinib, and midostaurin have improved patient outlook (7). These have been shown to produce clinical responses, but they are often transient and subject to resistance mechanisms (8). Overall, current standard of care regarding chemotherapeutics is limited, with prognosis remaining poor for many patients, warranting the need for detailed biological understanding of disease pathogenesis to develop future therapies (9). A deeper understanding of the molecular mechanisms underlying AML pathogenesis may offer new avenues to successful targeted therapies.

The heterogeneity of immune response between patients has made it difficult to characterize the landscape of AML tumor microenvironment. Several components within the tumor microenvironment interact to reprogram tumor initiation, progression, and response to therapies. Malignancies like AML deploy a plethora of immune escape mechanisms, including evasion of immunosurveillance systems, whereby cancer cells are able to avoid destruction (10). Such escape mechanisms in AML include aberrant expression of immune checkpoint receptors, deregulation of tumor necrosis receptor families and ligands which regulate T cell and natural killer cell responses, and immunomodulatory enzymes which abrogate anti-tumoral T cell effector function (11, 12).

Indoleamine 2,3-dioxygenase (IDO) is a haem-containing, immunomodulatory enzyme that catalyzes the oxidative cleavage of the indole ring of the essential amino acid tryptophan (Trp), producing immunoregulatory metabolites, collectively known as kynurenines [Kyn(s)] (13). IDO was originally described to promote immune tolerance in mammalian pregnancy, chronic infection, autoimmunity and allergic inflammation, but it has now become evident that it plays an important role in suppressing antitumour responses, and promoting tumor resistance (14). Some of the Kyn metabolites of tryptophan catabolism are known to inhibit T cell proliferation by arresting the cell cycle and inducing apoptosis (15). Direct action of Kyn has been associated with the generation of regulatory T cells (Tregs) (15). Natural Tregs, characterized by their constitutive expression of surface markers, such as CD25, CTLA-4, and the forkhead box P3 (FoxP3) transcription factor, are able to aid evasion of immune surveillance and suppress cytotoxic CD8⁺ T cell proliferation, favoring tumor progression (16).

Defining the tumor microenvironment and immune escape mechanisms of AML will support the development of new therapeutic strategies. IDO is of particular interest in this regard as studies report IDO expression in both bone marrow and in peripheral blood AML blasts, but the role it plays in disease progression remains unclear (12). Research aimed at defining tumor resistance and identifying immunosuppressive mechanisms in AML is expanding, but there is a lack of consolidated review that aims to systematically decipher the role of this immunosuppressive enzyme. Therefore, this study will review the role of IDO-driven Treg function in AML and evaluate the current body of evidence implicating IDO in AML pathogenesis, to determine its indication in prognosis and as a potential therapeutic opportunity.

A systematic review of literature was performed to identify the role of IDO AML progression. The review was conducted according to the guidelines outlined in the PRISMA statement (Supplementary Table 1) (17).

Of 190 papers identified through PubMed and Scopus search, 108 met the inclusion criteria once duplicates and non-English papers were removed. When full texts were screened, a final 20 studies were found to be suitable for inclusion in the study. The reasons for manuscript exclusions were those that had no mention of both IDO and AML (n = 28), manuscripts that did not tackle the role of IDO in AML (n = 20), those with unavailable abstracts or only abstracts were available (n = 20), review articles (n = 19), books (n = 2), those based on case studies alone (n = 3), and those focused on other types of cancer aside from AML (n = 1). A flow diagram detailing the screening process used to determine the studies suitable for inclusion in the systematic review is shown in Figure 1.

IDO has been shown to play an important role in the maintenance of immune tolerance, whilst also being implicated in several autoimmune inflammatory diseases, chronic infections and cancers (37). Its activity is majorly attributed to the enzymatic catabolism of the essential amino acid Trp, producing metabolites which arrest T cell proliferation, induce T cell apoptosis, and generate Tregs, producing an immune suppressive environment able to evade immunosurveillance systems. In AML, there is little biological understanding of disease pathogenesis resulting in high mortality rates despite efforts to refine current techniques and the development of targeted therapies. IDO and its Treg-inducing environment are not well-characterized in AML and so the clinical outlook for patients expressing IDO remains unclear.

This systematic review evaluates current knowledge on the role of IDO in AML, aiming to decipher its contribution to disease progression, prognosis, and potential therapeutic interventions. It is clear from the majority of the studies that IDO can be expressed on mononuclear cells of the peripheral blood, bone-marrow derived mesenchymal cells as well as leukemic blasts themselves (12, 18, 19, 21, 23–26, 31). Methodologies across these studies were consistent, focusing of peripheral blood and bone marrow, and utilizing PCR techniques. Only mesenchymal cell IDO expression was identified using bone marrow-derived samples with immunohistochemistry. This consistency across methodologies and samples used suggests that the expression of IDO is variable across multiple cell types. However, there occurs to be some discrepancy whether IDO is constitutively expressed (21, 24, 31) or induced (19, 21, 31). Consensus stands that whilst a subset of patients may have constitutive expression of IDO, others require it to be induced. These findings indicate an inter-patient variability in IDO expression in AML and that further studies should assess expression on a case-by-case basis to overcome the discrepancy. It also suggests that constitutive expression of IDO is not necessary to the progression of AML, but that it may favor progression.

In this review, a controversy was also highlighted regarding IDO expression relating to FAB subtypes. Some of the studies suggested that higher IDO expression is observed in the FAB-M5 subtype (19, 30, 31). The FAB-M5 subtype is associated with leukemia derived from a myelomonocytic lineage (38), and as aforementioned, IDO expression on mononuclear cells, including monocytes and monocyte-derived DCs has been found. As FAB-M5 is leukemia of monocytic origin and IDO can be expressed by these cells, it could account for this correlation. Contrasting this, four of the studies suggested that there was no significant difference in IDO expression between the FAB subtypes (20, 21, 25, 28). The methodologies used in these studies and those identifying a difference in IDO expression between subtypes were consistent. However, the sample size of those finding no difference were typically much larger (n = 650), providing more reliable statistical data. Accordingly, FAB subtypes should not be used to classify the likelihood of IDO expression.

AML mortality rates remain high despite attempts to refine treatment and develop more targeted techniques. Prognostic markers stratifying disease progression, such as the FLT3 mutation, have thus far only had limited effects on patient outcomes. Among the reviewed articles, seven studies assessed IDO expression prognosis (12, 20, 23, 25, 27, 28, 30). All associated high expression of IDO with poor prognosis, with some finding lower OS and CR attainment in IDO-positive groups. Across some of the studies, the value of IDO mRNA and Kyn levels as prognostic factors were assessed, finding that both hold potential as prognostic markers in AML (20, 25, 27, 29). Measurement of Kyn levels in patient samples are relatively easier than IDO mRNA measurements as this can be done in serum. However, a combination of both is likely to be most beneficial as a prognostic marker.

The Kyn metabolites produced by Trp cleavage can support Treg generation. The presence of Treg within the tumor environment is known to promote the development and progression of tumors by curtailing effective anti-tumor immune response, denoting poor survival in various types of cancers (39). Eight of the studies in the present review attempted to establish the role of Tregs in relation to IDO expression in AML (12, 18, 19, 24, 26, 32, 35, 36). These studies agreed that a positive correlation exists between IDO expression and Treg induction (12, 18, 19, 24, 26, 35), suggesting that in the case of AML, the immune suppressive environment may be resultant of the Treg accumulation induced by the expression of IDO. The most common CD4 Treg markers used across the studies were FoxP3 and CD25. However, they may both be transiently expressed by activated T cells in human and may not represent bonafide Tregs (40). CD127, is consistently expressed in high levels in activated T cells, but low or absent in Tregs, and therefore identification of CD25⁺FoxP3⁺CD127^low/− CD4 T cells may be more indicative of a Treg phenotype (41).

Pharmacological inhibition of IDO has been shown to break maternal immune tolerance in pregnancy, spontaneously aborting fetus, and precipitate autoimmunity in susceptible strains in mice (42, 43). Such findings provided proof-of-concept that breaking tolerance to tumor antigens in cancers where IDO is used as a route to immune escape may hold promise for cancer immunotherapy. Several studies included in this review also used the IDO inhibitor 1-MT in vitro. In samples from AML patients, 1-MT reduces Treg induction and Kyn levels, and increases effector T cell proliferation in vitro (23, 24, 26, 31, 32, 35). The 1-MT-D isoform, Indoximod, is an IDO inhibitor which exhibits a range of oral bioavailability across species, a good safety profile and the ability to slow tumor growth in preclinical studies (44, 45). In clinical trials, Indoximod has been tested as a monotherapy and in combination with chemotherapy agents where efficacy has been identified but primary endpoints have failed to be achieved, resulting in the termination of trials (45). Other IDO inhibitors that were not amongst the studies of this review, including the more potent Epacadostat, have also been trialed as therapies in many cancers but they lacked distinct objective responses (46). Preclinically, it has been suggested that a combination of IDO with checkpoint inhibitors may produce a synergistic effect that restores T cell deficits with cancer (45, 47). The ECHO 301 trial is the latest development exploring this clinically, testing Epacadostat in combination with anti-PD-1 antibody pembrolizumab in metastatic melanoma patients. This trial failed to demonstrate improved progression free survival and OS in the treatment combination compared to pembrolizumab monotherapy and thus terminated early (48).

Failure of the clinical trials to date raises questions as to where IDO inhibition could be incorporated into AML therapy, but it is important to note that although trials have taken place in other cancer types, few have focused on IDO inhibition in AML. Recently, a trial combining Indoximod with standard (7 + 3) chemotherapy was conducted in AML patients (NCT02835729) which has yet to report any findings (49). Results of this study could provide valuable insights into the rationale for continuation of trials investigating the clinical relevance of IDO inhibition therapies in AML. Nevertheless, early-mortality of AML patients with the highest IDO expression may highly likely benefit from IDO-inhibitor drugs as a component of their treatment regimen (22).

The present review is subject to limitations as only English papers could be included and a meta-analysis was not suitable due to the heterogeneity of data. It is also acknowledged that many of the papers came from the same authors and research groups/institutions therefore findings potentially could be subject to bias. The papers included in this review investigated IDO with no delineation of its isoforms or focused entirely on IDO1. It must be noted that IDO has two isoforms, IDO1 and IDO2, which exhibit high sequence homology but exhibit kinetic, functional and expression differences, with IDO2 expression limited to subsets including antigen-presenting cells (50). Current evidence suggests that IDO2 may serve an important role within immune control by supporting IDO1-dependent Treg function as part of the adaptive immune system (51). Therefore, it is important that the contribution of IDO2 in AML pathogenesis relating to Treg abundance and prognosis is explored in further studies.

To the best of our knowledge, this is the first systematic review highlighting the importance of IDO in AML progression and prognosis. IDO can be expressed in various cell types involved in AML pathogenesis and this expression can be either inducible or constitutive, differing between patients. The review identified the potential of IDO and Kyn levels as prospective prognostic markers as IDO expression correlates with worse patient outcomes. However, more robust CD4 Treg (CD4⁺CD25⁺CD127^low/−) markers may be required to delineate the contribution of bonafide Treg within the tumor microenvironment where IDO activity is functionally relevant. Furthermore, the role of other similar immunosuppressive enzymes, such as Tryptophan 2,3-dioxygenase may offer novel insights into the contribution of immunomodulatory enzymatic cascade in AML pathogenesis and prognosis.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

GW performed the search, screening and data extraction, interpreted the results, and wrote the manuscript. PK contributed to the results interpretation and writing the manuscript. LD designed the study, checked the accuracy of the data extraction, and contributed to the results interpretation and writing the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.