Immunomodulatory Role of Trp-Kyn-AhR Pathway

The Trp-Kyn-AhR pathway, in which the essential amino acid tryptophan is converted to Kyn and other secondary metabolites, is the primary route of tryptophan catabolism (19). Three enzymes catalyze the rate-limiting step of tryptophan catabolism to kynurenines: IDO1, IDO2, and TDO. The enzyme kynureninase hydrolyzes 3-hydroxykynurenine to 3-hydroxyanthranilic acid in the production of nicotinamide adenine dinucleotide. Regulation of these enzymes is notably divergent, with IDO1 being influenced by the interplay of IFNγ and interleukin (IL)-6 as compared with TDO being regulated by tryptophan, cholesterols, and prostaglandin E2 (20,21). The regulation and role of IDO2 is uncertain. Immunosuppression associated with Kyn was first described in experiments that demonstrated increased tryptophan catabolism limits allogeneic fetal rejection in mice (22).

In tumors, IDO1 is expressed by stromal cells of the TME and is induced by IFNγ as a result of CD8+ T cell infiltration and activation of other immunosuppressive pathways (7,23,24). TDO is ectopically expressed by tumor cells in certain malignancies (25). Kynurenines act as potent agonists of AhR, a ligand-gated transcription factor that is expressed in many immune cells and mediates a wide range of immunomodulatory effects (26,27). Elevated IDO1 and TDO activity and Kyn levels are associated with increased tumor grade and poor prognosis in many cancers. (28).

Several mechanisms have been proposed to explain the role of the Trp-Kyn-AhR pathway in tumor-associated immunosuppression. T cells are exquisitely sensitive to local depletion of tryptophan in which low tryptophan levels suppress mTORC pathways and activate general control nondepressible 2 (GCN2) kinase leading to cell cycle arrest and anergy of infiltrating T cells via eIF-2-dependent pathways (Figure 1) (29–31). However, recent studies have questioned the significance of this mechanism (32,33). Accumulation of kynurenines induce effector T cell arrest and lead to binding of AhR. This results in nuclear translocation and promotion of FoxP3 transcripts and IL-10, eventually producing regulatory T cell populations (18,34–37). In-vitro studies of AhR-deficient lung dendritic cells demonstrate failure to promote Treg development and an increase Th2 cell differentiation and pro-inflammatory responses to allergen exposure (38). AhR suppresses innate immunogenicity of antigen presenting cells and promotes IL-10 production by natural killer cells (Figure 2) (39–41). In addition, the Kyn-AhR interaction has been shown to upregulate PD-1 expression by CD8+ T cells via transcellular signaling mechanism in the tumor microenvironment (42).

Open in a separate window

Figure 1.

Tryptophan depletion-dependent signaling. Depletion of tryptophan suppresses activity in the mTORC1 signaling pathway, leading to autophagy in T cells, and releases GCN2-mediated phosphorylation of eIF-2, inducing cell cycle arrest and death in T cells.

Open in a separate window

Figure 2.

IDO1-Kynurenine-AhR signaling in TME immunosuppression. A, IDO1 in tumor cells, dendritic cells, and fibroblasts. TDO in hepatocytes are the rate-limiting enzymes in the conversion of tryptophan to Kynurenine and Kynurenine derivatives. Kynurenine binds to and activates the AhR, a ligand-activated transcription factor, in regulatory T cells, NK cells and dendritic cells. B, Activation and nuclear translocation of the AhR (1) in dendritic cells induces synthesis and release of IL-10 and inhibits IFNβ signaling, (2) in NK cells induces synthesis and release of IL-10 and IFNγ, and (3) in Tregs promotes Treg development. C, Tregs and IL-10 promote immunosuppression within the TME, whereas inhibition of IFNβ by AhR releases regulation of immunosuppression from inhibitory IFNβ signaling. In addition, both IL-10 and IFNγ promote IDO1 activity, establishing a positive feedback loop for IDO1-Kynurenine-AhR signaling.

Prominent Ido1/TDO Inhibitors and Trp-Kyn Pathway Inhibitors in Clinical Development

Several biochemical strategies exist to inhibit the Trp-Kyn-AhR pathway. IDO1 knockout mice demonstrate no clinical phenotype, in contrast to the inflammatory phenotype observed for knockouts of the immune checkpoints CTLA-4 and PD-1, and thus IDO1 inhibitors have predominantly been used in combination with other treatment modalities (43,44). Selective-IDO1 enzyme inhibitors such as epacadostat, NLG-919, and BMS-986205 either compete with tryptophan for the catalytic site of IDO1 or bind the enzyme with very high affinity (44–47). In contrast, the tryptophan mimetic indoximod appears to have pleiotropic effects on downstream Kyn-AhR pathway signaling and has been shown to relieve immunosuppressive signaling normally induced by tryptophan depletion (48,49). AhR inhibitors and recombinant kynureninase have more recently entered clinical development and will be discussed below.

A primary pharmacodynamic measure reported for selective-IDO1 inhibitors in clinical trials was reduction in peripheral blood Kyn levels. Initial peripheral blood Kyn suppression data demonstrated approximately 50% reduction suggesting other enzymes contribute to the production of systemic kynurenine, such as TDO. To date, assessment of intra-tumoral Kyn has not been consistently collected or reported in clinical trials (50,51). Figure 3 describes the prominent IDO, TDO inhibitors and Trp-Kyn pathway inhibitors currently in clinical development.

Open in a separate window

Figure 3.

Trp-Kyn pathway inhibitors in current or prior clinical development

IDO1, TDO and Trp-Kyn-AhR Inhibition in Combination Treatment

Association between the Trp-Kyn-AhR pathway and PD-1/L1 was suggested by the observation that both pathways are induced by IFNγ signaling in the tumor microenvironment (7,14). Indeed, across 30 human solid tumors from The Cancer Genome Atlas (TCGA) database, we have observed that the gene expression of IDO1 was strongly correlated with the expression of PD1 across increasing level of IFNγ responsive gene expression from non-T cell-inflamed to highly T cell-inflamed tumors (Figure 4A). In contrast, expression of IDO2, TDO2, KYNU, AHR and GCN2 (alias EIF2AK4) does not appear to correlate with PD1 expression or demonstrate IFNγ responsiveness on a transcriptional level as strongly as IDO1 (Figure 4B). Despite early observations for lack of monotherapy activity of selective-IDO1 inhibitors (52) combination strategies utilizing IDO1 inhibitors were quickly advanced. Indeed, IDO1 and PD-1/L1 inhibitor combinations appeared to show great promise in early phase clinical trials across multiple tumor types (Suppl Tables 1, 2).

Open in a separate window

Figure 4.

Expression of PD1 is positively correlated with immunotherapy relevant target genes across solid tumors from TCGA

. (A) Heatmap of Pearson’s product-moment correlation coefficient r between PD1 and immune target genes by tumor type. Immune target genes were separated into those strongly or less correlated with PD1 expression. Methods: level 3 RNA-Seq data (release date February 4, 2015) were downloaded for 30 solid tumor types from TCGA and processed as previously described (87). Acute myeloid leukemia, diffuse large B-cell lymphoma, and thymoma were excluded because of high tumor intrinsic immune cell transcripts. Skin cutaneous melanoma had both primary and metastatic samples available, whereas the other 29 cancers had only primary tumors available. A non-comprehensive list of 171 immune molecules representative of the interactions between tumor cells and immune cells in the tumor microenvironment were selected and correlated with PD1 (alias PDCD1) gene expression. For each tumor type, Pearson’s r was computed between each immune molecule and PD1 and used for clustering the genes by hierarchical unsupervised clustering with Euclidean distance. Two distinct groups are shown, consisting of (1) strongly correlated genes and (2) less correlated genes. (B and C) Correlation plots of PD1 vs IDO1, IDO2, KYNU, TDO2, AHR and GCN2 (alias EIF2AK4) (highlighted in red in A) in (B) metastatic melanoma and (C) NSCLC. Patients were categorized into T cell-inflamed (red), non-T cell-inflamed (blue) and intermediate groups using a defined T cell-inflamed gene signature (87). Each data point represents one patient. NSCLC = non-small cell lung carcinoma.

A substantial literature also supports the potential utility of inhibition of the IDO pathway in conjunction with other anti-cancer modalities. Studies of IDO pathway blockade with radiation, chemotherapy and tumor vaccines suggest an improvement relative to those treatments alone (53,54). Several clinical trials evaluating combinations across these modalities are on-going (Suppl Table 1).

Epacadostat

Epacadostat, a competitive, selective inhibitor of IDO1, reached the most advanced stage of development, with the recent early termination of ECHO-301/Keynote-252; a phase III clinical trial in combination with pembrolizumab in metastatic melanoma. At median follow up of 14 months, patients treated with epacadostat plus pembrolizumab demonstrated a progression free survival of 4.7 months versus 4.9 months in those treated with pembrolizumab plus placebo (HR=1.00; CI, 0.83–1.21; P=0.517.) The overall response rate was 34.2% versus 31.5% in the epacadostat plus pembrolizumab and placebo plus pembrolizumab groups respectively. Treatment related adverse events occurred in 79.3% of patients receiving epacadostat plus pembrolizumab versus 81.0% receiving placebo plus pembrolizumab and Grade >3 treatment-related adverse events occurred in 21.8% versus 17.0%, respectively (55). These negative results were unexpected as preclinical studies and early phase clinical trials of this combination in as many as 14 different solid tumors showed encouraging results (Suppl Table 1). Multiple hypotheses have been advanced to explain the seeming discrepancy between early phase clinical trial success of epacadostat and the failure of the late phase ECHO-301 trial. A non-exhaustive list of possible explanations to differentiate early vs late phase results could include differences between the treatment populations, inappropriately low dosing of epacadostat and incomplete suppression of intra-tumoral Kyn.

Regarding the patient populations for these studies, the patient characteristics appeared to be relatively similar from the early (ECHO-202) to late phase (ECHO-301) melanoma studies across multiple variables including but not limited to performance status of zero (77 vs 76%), M1c staging (55 vs 61%), elevated lactate dehydrogenase (37 vs 32%) and no prior therapy (71 vs 89%), respectively (55,56). As IDO1 expression is intimately linked to IFNγ gene expression, the baseline quality of the T cell-inflamed tumor microenvironment may be of relevance to the results. To date, this is not well characterized in either study. In ECHO-301 IDO1 expression was not an inclusion criterion (though PD-L1 expression was a randomization stratification factor) with only PD-L1 status as a surrogate from the early phase studies. Given the seeming complete lack of activity between epacadostat and placebo in ECHO-301, some have wondered whether further study should be given to the patients treated in the early phase study. Despite the lack of obvious clinical differentiators, perhaps the tumors from these patients were disproportionately T cell-inflamed and thus much more likely to respond to pembrolizumab and possibly epacadostat.

Other less commonly controlled for differences of potential clinical significance could exist between the early to late phase trial populations also however. One of these would be to note that the early phase trial was conducted at a select number of sites in the United States while the late phase study was predominately international (Australia, Europe, Asia, South America) with a lesser accrual in the United States. It has been observed in previous trials of melanoma that outcomes appeared to be substantially different between these populations in the phase III setting (57). Consideration might be given to whether dietary or environmental exposures could be variable in these different localities especially given the canonical role of AhR as a xenobiotic sensor that is responsive to signals from the gut microbiome (18). With an evolving literature supporting the microbiome as a potential influencer of PD1 antibody response (58), focus on diet, medication use and microbiome contents may be important variables to track in clinical trials moving forward.

Regarding dosing of epacadostat, 100 mg was taken forward out of the phase I study despite higher dose levels being tolerable and no maximum tolerated dose being established (59). This dose was chosen by Incyte as it was deemed that maximal inhibition of IDO1 activity was observed at doses of ≥100 mg with a relative plateau in decrease of kynurenine level in peripheral blood at higher doses. This point of the most appropriate dose is debated however particularly given that published modeling of IDO inhibition only reaches approximately 50–70% at the 100 mg dose (60). It is somewhat notable that an early phase study of epacadostat with nivolumab was simultaneously pursued using 300 mg of epacadostat (61) and published modeling implies a higher likelihood of IDO1 inhibition at this dose (60). The optimal dose of epacadostat continues to be explored in on-going clinical trials.

Perhaps of most relevance to the failure of ECHO-301 however is the open question of intra-tumoral pharmacodynamics. While peripheral blood monitoring of kynurenine was reported for epacadostat in phase I, to date no intra-tumoral data has been released from any clinical trial. Given the lack of consensus surrounding whether limiting tryptophan depletion versus suppression of Kyn acts as a primary mechanism of immunosuppression, this data is essential to inform the field surrounding next steps. Even very low levels of canonical AhR ligands, such as Kyn and dioxin, can activate AhR associated gene expression and recent preclinical studies have suggested kynureninase and direct AhR inhibitors have higher potency relative to IDO1 selective inhibitors (62–64). If suppression of Kyn is indeed the dominate mechanism it is very possible that IDO1 inhibition alone may be inadequate to drive intra-tumoral levels consistently low enough to alleviate the immunosuppressive effects of Kyn-activated AhR, including production of IL-10 and suppression of type I IFN (41).

Other Selective-IDO1 Inhibitors

Pivotal studies of BMS-986205, an irreversible IDO1 inhibitor developed by Flexus Biosciences and Bristol Meyers Squib, have been predominately scaled back in wake of epacadostat’s late-stage trial failure though a randomized study in bladder cancer is still planned. Relative to epacadostat, BMS-986205 demonstrates higher potency based on IC50 in IDO1 expressing cell lines. In contrast to the lack of such data for epacadostat, analysis of 39 paired pre- vs on-treatment tumor samples across various tumor types from the Phase I trial of BMS-986205 plus nivolumab demonstrated decreased Kyn levels (and mostly near zero levels on-treatment) and increased the percentage of proliferating CD8⁺ T cells (65).

Genentech has recently terminated rights to NewLink’s NLG-919 (navoximod/GDC-0919), another selective inhibitor of IDO1.

Tryptophan Mimetics

Indoximod, the D-enantiomer of 1-methyl-tryptophan, has demonstrated inhibition of the IDO pathway as a tryptophan mimetic. Indoximod limits IDO-mediated immunosuppression by at least two mechanisms including 1) serving as an artificial Trp-sufficiency signal that prevents activation of GCN2 and inhibition of mTORC1 and 2) modulation of AhR-dependent transcriptional activity (66). Indoximod increased activity and proliferation of CD8+ T cells by limiting tryptophan-depletion mediated mTORC1 suppression (66). mTORC1 activation has been associated with ICOS expression, a T cell coregulatory receptor seen on tumor-infiltrating T cells that has been associated with clinical response (67). In an AhR-dependent manner, indoximod was shown to stimulate CD4+ T cell differentiation to Th17+ helper T cells, inhibit FoxP3 T regulatory cells and downregulate expression of IDO in dendritic cells (66).

Data from a single-arm phase 2 trial of indoximod plus anti-PD1 in advanced melanoma achieved an ORR of 56% and CR in 19% with low rates of high-grade immune-related adverse events (68). Multiple phase II and III trials combining indoximod with other current modalities of treatment, including chemotherapy, cancer vaccines, and checkpoint immunotherapy are ongoing (Suppl Tables 1, 2).

Dual IDO1/TDO Inhibitors

Analysis of tumor and immune cells by immunohistochemistry revealed differences in the expression of IDO1 and TDO among tumor types, suggesting the potential for a possible advantage with dual IDO and TDO inhibitors in certain tumors (69). Dual IDO/TDO inhibitors such as RG70099 decrease serum Kyn levels by approximately 90%. IOM-E and IOM-D are selective-IDO1 and dual IDO1 and TDO inhibitors, respectively. Preclinical studies have revealed a promising pharmacokinetic profile. Significant in vivo efficacy was observed in mouse pancreatic adenocarcinoma cells treated with IOM-E, the selective-IDO1 inhibitor, in combination with gemcitabine and abraxane. Particular efficacy in preclinical lung cancer models has also been seen (70). Several companies have disclosed pre-clinical programs surrounding the development of dual IDO/TDO inhibitors though the current status of these programs is in flux in wake of ECHO-301. It is worth pointing out however that in murine models, complete IDO/TDO inhibition results in significant alteration of Trp metabolism which has raised concern over potential neurologic toxicity (seizures) from dual IDO1/TDO inhibition (71).

AhR Inhibitors

Inhibition of AhR signaling with small molecule antagonists interferes with the downstream immunomodulatory effects irrespective of the source of Kyn production. Early inhibitors in this class have been shown to block nuclear translocation of AhR and enhanced production of IFNγ, TNF, IL-2 and reduction in tumor associated M2-like macrophages. In mouse models, AhR inhibitors have demonstrated activity as monotherapy, a notable contrast to IDO1 inhibitors. Furthermore, enhanced activity is seen with AhRi combination with anti-PD-1 (72). Several biotechnology companies including Hercules Pharmaceuticals, Ideaya Biosciences and KYN Therapeutics have disclosed the development of AhR inhibitors.

Prior to the failure of ECHO-301 and near complete re-work of the manuscript, medical writing and editorial support were provided by Jeremy Kennard, PhD, and Shannon Davis of Infusion Communications, Middletown, CT, with funding by Incyte Corporation, Wlimington, DE.

Funding: JJL is supported by Department of Defense Career Development Award (W81XWH-17–1-0265), The National Cancer Institute (P30CA014599–43S), the Arthur J Schreiner Family Melanoma Research Fund, J. Edward Mahoney Foundation Research Fund and Brush Family Immunotherapy Fund as well as support from Center for Research Informatics of The University of Chicago Biological Science Division and The Institute for Translational Medicine/CTSA (NIH UL1 RR024999).