Indoleamine 2,3-dioxygenase 1-mediated immune suppressive status is positively associated with brain metastasis in patients with non-small cell lung cancer

Abstract

Background

Indoleamine 2,3-dioxygenase (IDO1) activity, measured by kynurenine/tryptophan (K:T) ratio, is known for its association with distant metastasis and overall survival (OS) in patients with non-small cell lung cancer (NSCLC). Here, we aimed to examine whether IDO1 activity is correlated with OS in NSCLC patients with brain metastasis (Bramet) and has negative effect on modulating the anti-tumor functions of immune cells.

Methods

This study was a part of a prospective clinical trial in circulating biomarkers. Blood or tissues from eligible participants were collected for measurement of kynurenine, tryptophan, immune cell subtype, scRNA-seq analysis, and untargeted metabolomics analysis.

Results

A total of 195 patients were enrolled. The median kynurenine to tryptophan (K:T) ratio was 0.18, with consistent values observed among patients with NSCLC Bramet and those without (0.18 and 0.11, respectively). Notably, student's t-test analysis revealed significantly higher kynurenine concentrations in stage IV patients compared to those in stage I (2.3 vs 1.7 µM, P < 0.001). In patients with Bramet, both kynurenine concentrations and K:T ratios were significantly elevated in comparison with those of extra-cerebral metastasis (2.7 vs 1.9 µM, P < 0.001; 0.12 vs 0.095, P = 0.028; respectively). Single-cell analysis further validated a high level of IDO1 expression in stage IV tumors or Bramet lesions, particularly in macrophages, regulated by chemokines such as CXCL11. Additionally, K:T ratios exhibited significant associations with Treg cell percentages and OS in patients with Bramet (P = 0.039). Treatment with kynurenine led to the upregulation of immune-suppressive molecules, including PD-1, in T cells. Finally, untargeted metabolomics analysis further identified that, apart from the IDO1 metabolic pathway, other metabolites, such as those involved in phospholipid pathways, were also implicated in Bramet.

Conclusion

IDO1 metabolites may play immune-suppressive roles in NSCLC patients with Bramet.

1. Introduction

Approximately 20 % of patients with non-small cell lung cancer (NSCLC) have brain metastasis (Bramet) at the initial diagnosis, and up to 50 % of these patients develop Bramet during disease development.1, 2, 3 Bramet is responsible for about 40 % of lung cancer-related death, and the median overall survival (OS) time was 3.0–14.8 months.4, 5, 6 An early intervention of patients with possible progression of Bramet, monitored by blood biomarkers, can help tailor the personalized treatment regimen and improve survival in these patients.⁷

At the time of intra-cerebral metastasis, cancer cells undergo metabolic rewiring, leading to metabolic heterogeneity and plasticity in the tumor immune microenvironment (TIME).^8,9 Indoleamine 2,3-dioxygenase 1 (IDO1) is a rate-limiting enzyme that metabolizes tryptophan (Trp) to kynurenine (Kyn).^10,11 The release of Kyn can promote regulatory T cells (Tregs) differentiation¹² and attenuate the immune responses of effector T cells and natural killer (NK) cells,^13,14which cultivates an immune-suppressive TIME. IDO1 was found to be upregulated in Bramet of colorectal carcinoma and associated with tumor progression of lung cancer in TIME.^15,16 In the systemic tumor immune environment (STIE), the activity of IDO1 can be monitored by the ratio between Kyn and Trp (K:T ratio) in the plasma and serve as a biomarker in tumor progression.^17,18 For example, the K:T ratio increased in the serum or plasma of patients with malignant melanoma and gynecological cancers.^19,20 Additionally, the plasma or serum concentrations of Kyn were positively associated with the disease progression and the risk of lymph node metastasis in hepatocellular carcinoma²¹ and lung cancer.^22,23 Furthermore, the circulating concentrations of Kyn were closely associated with the immune suppressive cell functions in lung cancer patients.²⁴ The findings hinted at the possible uses of assessing circulating IDO1 activity to track tumor advancement and the immune condition of the host. However, the alterations in IDO1-mediated immune activity and its correlation with Bramet in lung cancer patients remain ambiguous.

In this study, we hypothesized that IDO1 activity in circulating blood could impair immune cell functionality in patients with NSCLC Bramet. Our specific objectives were twofold: firstly, to explore the correlation between plasma levels of Kyn, Trp, mRNA expression of IDO1, with tumor stages and the occurrence of Bramet. Secondly, we aimed to assess the impact of IDO1 metabolites on the anti-tumor capabilities of immune cells by analyzing changes in immune subtypes in NSCLC patients and the expression of T cell dysfunctional markers. This investigation sheds light on potential mechanisms underlying immune dysregulation in NSCLC patients with Bramet and may offer insights for the development of targeted therapeutic.

2. Materials and methods

2.1. Patient population

This was part of a multicenter prospective observational study of circulating immune biomarkers for prediction and prognosis (NCT05061342), started from June 27, 2019. The inclusion criteria of cancer patients were: 1) 18 years of age and older; 2) scheduled to receive anticancer therapy in our center; 3) performance status of Eastern Cooperative Oncology Group (ECOG) 0, 1, 2, or 3; 4) able to understand the questionnaire. Exclusion criteria of enrollment: those who have limited ability to complete the survey questionnaires of quality of life will be excluded. Newly diagnosed or recurrent NSCLC patients were eligible=. Clinical patient factors were collected, including age, histology, clinical stage, gender, smoking history, body weight, and ECOG score. Bramet patients were identified by the magnetic resonance imaging (MRI) of brain with contrast and computerized tomography with IV contrast in cases that can not have MRI. Healthy individuals of 18 years of age and older without a history of cancer except for cured skin cancer served as controls.

2.2. Blood sample and platelet poor plasma preparation

A 10 mL of peripheral venous blood was collected in EDTA-anticoagulant blood tubes from the enrolled patients. Blood samples were centrifuged at 1600 g, 4 °C for 10 min to separate the supernatant from the buffer coat layer. The supernatants were followed by a 10 min at 16,000 g to remove cell debris and small platelet completely. The platelet-poor plasma was stored at –80 °C until analysis. The buffy coats were stored in 90 % heat-inactivated fetal bovine serum (HI-FBS, Gibco, #10270106) + 10 % DMSO (Gibco, #D2650) with gradual cooling to –80 °C for analysis of immune cell subtypes.

2.3. Measurements of plasma Trp and Kyn

Trp and Kyn concentrations in platelet-poor plasma were measured using high-performance liquid chromatography as previously described with minor modifications.²⁵ A total of 25 uL of plasma samples were diluted with equal volumes of 30 mmol/L NaAc (pH 4.0) and deproteinated with perchloric acid to 5 %. Then the mixture was vortexed and incubated in ice for 5 min, followed by centrifugation at 4500 rpm (RPM) for 5 min. The supernatant was then transferred to a filtered plate and centrifuge at 3200 RPM for 5 min. All supernatants were loaded onto reverse-phase 5 µm, 5/4.6 C18 column (Agilent) using an autosampler. Chromatography was performed as follows: mobile phase A (2.5 % acetonitrile in 15 mM NaAc pH4.0) and mobile phase B (acetonitrile) were delivered with solvent pumps (A) at the flow rate of 1.5 mL per minute. The column was flushed with 10 % mobile phase A + 90 % mobile phase B and then re-equilibrated with the initial mobile phase after every sample. Kyn was detected on a UV/vis channel at 360 nm, and Try was detected at 285 nm excitation and 365 nm emission (Agilent). Samples were quantified with external standards, and at least one quality control sample was randomly inserted into every 30 samples for reference. For quality control purposes, double-blinding duplicate testing verified the assay reproducibility to be over 95 %.

2.4. Measurements of immune cell subtypes

Single-cell suspensions were prepared from stored buffy coats. The buffy coat was treated with the red blood cell lysis buffer (BioLegend, #420301) and stained for Zombie Aqua™ Fixable dye (BioLegend, #423101). After blocking the cells with the Fc antibody, the cells were stained with the following antibodies CD8a, CD11b, CD45, CD4, CD19, CD56, CD3, CD25, and CD127. Or the cells were stained by CD8a, CD11b, CD45, CD4, CD19, CD56, and CD3, and then fixed and permeated, followed by staining with the intracellular antibody FOXP3 detected by a flow cytometry machine (BD LSR Fortessa flow cytometry or Aria III). The cell subtypes were defined as followings: Treg cell (CD4⁺CD25⁺CD127^- or CD4⁺Foxp3⁺), NK cell (CD56⁺CD3^-), NKT cell (CD56⁺CD3⁺), B cells (CD19⁺CD45⁺), total T cell (CD45⁺CD3⁺), CD4⁺ T cell (CD45⁺CD3⁺CD4⁺), CD8+ cell (CD45⁺CD3⁺CD8⁺) and monocyte (CD11b⁺). The percentage of immune cell subtypes was further analyzed using FlowJo software (TreeStar, V10). The anti-human antibodies used in this study purchased from BioLegend were listed in Supplemental Table 1.

2.5. In vitro T cell function assay

T cells were purified from the peripheral blood mononuclear cells (PBMCs) from healthy subjects and treated with Kyn for in vitro functional assays. Briefly, T cells were isolated with pan-T cell isolation kits (BioLegend, #480021) and then activated with CD3/CD28 kit ((IBA Lifesciences, #15623988) for T cell expansion (30 µL CD3/CD28 mixture per well, 1 × 10⁶ cells). The cells were cultured in T-cell media (RPMI 1640 supplemented with 10 % HI-FBS, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 µM beta-mercaptoethanol, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 10 mM HEPES, and 1 % penicillin-streptomycin), supplemented with 30 U/mL recombinant human-IL-2 (R&D system, #202-IL-050), unless otherwise stated. The cells were cultured for 48 hours and followed by Kyn treatment at a concentration of 0, 0.05, 0.1, 0.4, and 0.8 mM or from Bramet or not Bramet patients for 24 hours. The cells were then harvested and stained for the following antibodies: CD3, CD45, CD4, CD8, Zombie Aqua™ Fixable dye, programmed death-1 (PD-1), cytotoxic T lymphocyte antigen 4 (CTLA-4), lymphocyte activation gene 3 (LAG-3) and T cell immunoglobulin and mucin-containing gene 3 (Tim-3). The stained cells were analyzed by the NovoCyte Quanteon Flow Cytometer (Agilent). Data are representative of three independent experiments. The detailed information on these antibodies is listed in Supplemental Table 1.

2.6. Single-cell RNA and spatial transcriptome library construction

For single-cell RNA sequencing, fresh tumor tissues stored in tissue storage solution (Miltenyi Biotec) were sent to BGI genomics for single cell analysis on the Chromium platform (10 × Genomics) using the chromium next GEM single cell 3′ library construction kit, completed by the BGI genomics. Fresh tissues were embedded in optimal cutting compound media, cut into 10 µm sections. The 10 × Genomics Visium spatial gene expression reagent kits were used for library construction in the 10 × Genomics spatial RNAseq visium platform, completed by the BGI genomics.

2.7. Untargeted metabolomics analysis

Platelet poor plasma was used for metabolomics analysis. A 20 µL plasma sample was first mixed with 120 µL of 50 % methanol in ice. The mixture was vortexed, incubated at room temperature, and then stored at −20 °C overnight. The mixture was centrifuged at 4000 g for 20 min to collect the supernatant. 10 µL of each sample extraction was mixed to make the pooled quality control (QC) samples. Extraction was injected into the T3 column (Waters) using an autosampler and further analyzed using an ultra-performance liquid chromatography system (UPLC; SCIEX) with a tandem mass spectrometer (MS; SCIEX) in both negative electrospray ionization (ESI-) and positive electrospray ionization (ESI+) detection modes, which was completed by the LC-biotechnologies.

2.8. RNA extraction and quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Human lung cancer Bramet tissue and paired blood were collected from surgery patients from Queen Mary Hospital. Brain tissue was extracted to RNA collection using an RNA extraction kit (Roche). The RT-PCR kit (Roche) synthesized the cDNA for qPCR by a SYBR Green PCR Kit (Bio-Rad) on a 384 plate with LCM 480 Detector (Roche). The sequences of primers were listed in Supplemental Table 2.

2.9. Data and statistical consideration

IDO1 concentrations in different tumor stages and the expressions of immune cell dysfunctional markers under treatment with serial Kyn concentrations were compared using the One-way ANOVA test for multiple testing. In addition, box plots of IDO1 levels between Bramet and extra-cranial metastasis patients were compared using Student's t-test by GraphPad Prism (version 10.01). Descriptive statistics results [mean value and 95 % confidence interval (CI)] were shown.

OS was the primary outcome endpoint for this study, defined as the time (in months) from the first blood drawn to the date of death of any cause. The R package “survminer” was used to determine the cut-off point of the IDO1 metabolites by examining the optimal P-value. The patients were divided into high-level and low-level groups by the cut-off point. Log-rank test was used to compare the survival outcomes between two different groups. The relationship between the immune cell subtypes and IDO1 metabolites was analyzed with Pearson correlation. P < 0.05 was considered statistically significant unless otherwise specified.

Single-cell data analysis: Public datasets (GSE131907, GSM5645894, GSM3516662, and GSE123904) were downloaded from the GEO database. The subsequent data analysis was performed using the R package: Seurat (4.0.5). The harmony algorithm was used to remove the batch effect between different dataset sources or different tissue origins or different stages. Data were normalized using transforming the counts per million (CPM) divided by 100.0 and then adding 1.0 by log exponential transformation. Graph-based clustering method was used by the Louvain Clustering algorithm with the nearest neighbor type (K-NN). The cells were classified into the following major cell types: CD4⁺ T cells (CD3⁺CD8^-CD4⁺), CD8⁺ T cells (CD3⁺CD8⁺NKG7^-), NK cells (NKG7⁺), myeloid cells (CD14⁺), endothelial cells (PECAM1⁺VWF⁺), mesenchymal cells (COL1A1⁺DCN⁺), B cells (CD79A⁺), mast cells (TPSB2⁺TPSAB1⁺), epithelial (cancer) cells (EPCAM⁺CDH1⁺) and oligodendrocytes (PDGFRA⁺EDNRB⁺) according to the summary of the published paper. T-distributed stochastic neighbor embedding (t-SNE) or uniform manifold approximation and projection (UMAP) was used to visualize the single-cell analysis data. Post-Hoc Tests (Bonferroni correction) were used for multi-group comparison. The AUCell algorithm was used to score each cell for enrichment of specific transcriptomes.

Spatial transcriptome analysis was performed using the R package: Seurat (4.0.5). Raw spatial gene expression data were first loaded and quality controlled by filtering out spots with fewer than 200 genes and genes detected in fewer than 3 spots. The data were then normalized using the SCTransform method, which accounts for technical factors such as sequencing depth. Variable features were identified using the “vst” method, selecting the top 1000 genes. Dimensionality reduction was performed using principal component analysis (PCA) on the variable features, and the first 15 principal components were used for downstream analyses. t-SNE was applied for visualization of the data in two dimensions. Spatial clusters were identified using the SpatiallyVariableFeatures function with the “markvariogram” method, followed by graph-based clustering using the FindClusters function with a resolution of 0.8. Spatial gene expression of IDO1 patterns were visualized using the SpatialFeaturePlot function.

Metabolomics data processing: The liquid chromatography-mass spectrometry (LC-MS) raw data were transformed to the mzXML format through the R software toolboxes: XCMS, CAMERA, and metaX. Each ion was defined by the retention time, the mass-to-charge ratio (m/z), and the intensity. These metabolites were annotated using the in-house and public databases: Kyoto Encyclopedia of Genes and Genomes (KEGG) and Human Metabolome Database (HMDB). Further analysis of the data by Pareto scaling was performed through the R package “MetaboAnalystR”. Partial least-squares discriminant analysis (PLS-DA) was applied to compare the metabolic profile differences between patients. The variable importance of projection (VIP) was used to identify the significant metabolites. KEGG analysis was used to further annote these genes’ functions.

3. Results

3.1. Baseline patient characteristics

A total of 195 NSCLC patients were enrolled in this study, with stage distribution as the followings: stage I = 46, stage II = 19, stage III = 29, and stage IV = 101. The median age for all patients was 52 years (range: 21–88). Among these patients, 77 % had adenocarcinoma, and 17 % had squamous-cell carcinoma. Blood samples were collected at the timing when patients started new lines of treatment. These patients may be newly diagnosed (i.e., treatment-naïve patients) or recurrent (i.e., patients with previous lines of treatment). Among them, 58 patients were treatment-naïve at the time of the study. 49 patients (25 %) had previously received targeted therapy, while 64 patients (33 %) had chemotherapy before enrollment (Table 1).

Table 1.

Baseline patient characteristics.

Factors (n = 195)	No. (%)
Gender
Male	120 (61)
Female	75 (39)
Age, median (range), years	51(21–88)
Smoking history
Non-smoking	97 (50)
Current smoker	39 (20)
Ex-smoker	25 (13)
Unknown	34 (17)
ECOG at enrollment, median (range)	2 (0–3)
Stagea
I	46 (24)
II	19 (9)
III	29 (15)
IV	101 (52)
Histology
ADC	150 (77)
SCC	33 (17)
Unspecifiedb	12 (6)
Lines of treatment prior to blood sampling
0 (treatment-naive)	58 (30)
1	60 (31)
2	24 (12)
3	17 (9)
4	22 (11)
5	14 (7)
Antitumoral treatment prior to blood sampling
Targeted therapy	49 (25)
Chemotherapy	64 (33)
Surgery	6 (3)
RT	4 (2)
Immunotherapy	4 (2)
Unspecified	10 (5)
No	58 (30)

^aStage according to AJCC cancer staging manual, eighth edition.

^bMixed histology non-small cell lung cancer.

Abbreviations: ADC, adenocarcinoma; ECOG, Eastern Cooperative Oncology Group; RT, radiotheraoy; SCC, squamous cell carcinoma.

3.2. IDO1 metabolites and mRNA expression increased in advanced stages of NSCLC patients

The blood or tissues from eligible participants were measured for kynurenine, tryptophan, immune cell subtype, scRNA-seq analysis or in vitro validation of IDO1 metabolites’ functions (Fig. 1A). We initially measured the STIE IDO1-metabolites using the high performance liquid chromatography (HPLC) in both 195 NSCLC patients and 8 normal controls (Fig. 1B). The mean Kyn concentration in normal controls (n = 8) was 1.7 µM (95 % CI, 1.4–2.0 µM), and 1.9 µM (95 % CI, 1.8–2.0 µM) in NSCLC patients (n = 195), showing no significant difference (P = 0.71). However, Trp concentrations were significantly lower in NSCLC patients (23.5 µM; 95 % CI, 22.2–24.7 µM; P < 0.001) comparing to normal controls (36.5 µM; 95 % CI, 33.0–40.0 µM) (Fig. 1C and Supplemental Fig. 1A).

Fig. 1.

IDO1 metabolites and stage in patients with NSCLC. (A) Representative plots show design of the study. (B) Representative plots show detection peak of Trp and Kyn using HPLC. (C) The differences in IDO1 metabolites between NSCLC and normal controls (healthy) are shown in Kyn concentration, Trp concentration, and K:T ratio. Significance was tested with Student's t-test. (D) IDO1 metabolites in different stages (stage I, II, III, IV) are shown in Kyn concentration, Trp concentration, and K:T ratio. (E) Violin plots show that IDO1 expression [Log₂(CPM+1)] from stage I to stage IV in NSCLC patients. (F) Kaplan-Meier curves showing a significant correlation between the baseline levels of Kyn or K:T ratio and survival in stage IV patients. 95 % confidence intervals are indicated by the error bar. *, ^⁎⁎, ^⁎⁎⁎, and^⁎⁎⁎⁎ indicates P 〈 0.05, P < 0.01, P < 0.005, and P < 0.001, respectively. Conc, concentration; FACS, fluorescence activated cell sorting; HPLC, high performance liquid chromatography; Kyn, kynurenine; K:T ratio, kynurenine vs. tryptophan; Ns, not significantly different, P > 0.05; NSCLC, non-small cell lung cancer; Trp, Tryptophan.

We further identified the changes of IDO1 metabolites in NSCLC patients with different stages. The mean Kyn concentration was 1.3, 1.4, 1.9, and 2.3 µM for stages I, II, III, and IV, respectively. The concentrations of Kyn (2.3 µM; 95 % CI, 2.1–2.5 µM) in patients with stage IV were significantly increased compared with 1.3 µM in patients with stage I disease (95 % CI, 1.1–1.5 µM; P < 0.001). The Trp mean concentration was 24.3, 20.1, 25.0, and 23.3 µM in patients with stages I, II, III, and IV, respectively. The concentrations of Trp were not significantly different across the distinct stages of NSCLC. The K:T ratios were significantly higher in stage IV patients (0.10; 95 % CI, 0.08–0.12) than that in stage I patients (0.045; 95 % CI, 0.04–0.05, P = 0.001) (Fig. 1D, Supplemental Fig. 1B and C).

We further characterized the expression pattern of the IDO1 in different tumor tissues via the single cell RNA sequencing datasets. A total of 239,787 cells from 57 treatment naïve patient tumors (published data, n = 53; original data, n = 4) were selected (stage I, n = 22; stage II, n = 3; stage III, n = 4; stage IV, n = 28). Cells were clustered into different cell types as followings: CD4⁺ T cells, CD8⁺ T cells, NK cells, myeloid cells, endothelial cells, mesenchymal cells, B cells, mast cells, epithelial cells (tumor featured cells), and oligodendrocytes (Supplemental Fig. 2A). The expression of IDO1, indicated by the mean of log₂(CPM+1), in stage IV was 0.17, which was significantly higher than that in stage I (mean, 0.11), II (0.07), and III (0.05; P < 0.001) (Fig. 1E) and abundantly expressed in myeloid cells and epithetical cells in patients with stage IV NSCLC (Supplemental Fig. 2A).

The association between the survival of NSCLC patients in stage IV and Kyn concentrations and the K:T ratio were further analyzed to investigate the prognostic value of IDO1 metabolites. Notably, it was found that patients with relatively high levels of Kyn displayed worse survival (median OS, 11.0 months) than those with low Kyn concentrations (median OS, 17.1 months; P = 0.011; cut-off point = 1.35). The patients with a relatively high K:T ratio (median OS, 10.2 months) were associated with poor survival outcomes compared with those with a low K:T ratio (median OS, 17.1 months; cut-off point = 0.05, P = 0.005) (Fig. 1F). Univariate and multivariate Cox regression analysis indicated that Kyn was a significant risk factor for OS in stage IV patients (hazard ratio [HR], 1.92; 95 % CI, 1.10–3.35; P = 0.021). However, the concentrations of Trp and the K:T ratio were not associated with OS in stage IV patients (P > 0.05) (Table 2).

Table 2.

Univariate and multivariate Cox regression analysis of patient characteristics and IDO1 metabolites for overall survival among stage IV patients.

Variable	Univariate HR (95 % CI)	P-value	Multivariate HR (95 % CI)	P-value
Smoking
Non-smoking	Reference	Reference
Chronic smoker	0.63 (0.22–1.78)	0.383
Ex-smoker	0.64 (0.18–2.24)	0.486
Subtype
ADC	Reference	Reference
SCC	0.46 (0.06–3.44)	0.451
Other	0.85 (0.26–2.85)	0.796
EGFR mutant
No	Reference	Reference
Yes	0.89 (0.39–2.04)	0.789
PD-L1 expression, %	1.01 (0.98–1.03)	0.596
Previous Treatment
No	Reference	Reference
Chemotherapy	0.53 (0.07–4.09)	0.541
Targeted therapy	0.34 (0.04–2.72)	0.308
Others	0.18 (0.02–2.07)	0.168
Age, years	0.97 (0.94–1.01)	0.110
Gender
Male	Reference	Reference
Female	1.50 (0.71–3.16)	0.292
Kyn, µM	2.19 (1.25–3.83)	0.006	1.92 (1.10–3.35)	0.021
Trp, µM	0.99 (0.94–1.04)	0.638
K:T ratio, ln	1.99 (0.97–4.10)	0.061
Post-treatment
Surgery	Reference
Radiotherapy	4.42 (0.49–13.12)	0.07
Chemotherapy	-	–
Targeted therapy	4.93 (0.54–45.41)	0.159	6.11 (0.65–57.42)	0.114
Immunotherapy	5.91 (0.64–54.65)	0.117	5.26 (0.56–49.03)	0.145

The bold values mean the differences between these comparing groups are statistical significance, which P value < 0.05.

Abbreviations: ADC, adenocarcinoma; CI, confidence intervals; HR, hazard ratio; IDO1, indoleamine 2,3-dioxygenase; K:T ratio, Kynurenine vs. Tryptophan; Kyn, Kynurenine; SCC, squamous cell carcinoma; Trp, Tryptophan.

3.3. Metabolites of IDO1 and mRNA expression are elevated NSCLC patients with Bramet

When comparing patients with Bramet to those without, we found that Kyn concentrations were significantly higher in Bramet patients (2.7 µM; 95 % CI, 2.4–3.0 µM) than in those without Bramet (1.7 µM; 95 % CI, 1.5–1.8 µM). The K:T ratio was also significantly higher in Bramet patients (0.12; 95 % CI, 0.10–0.15) than in those without (0.08; 95 % CI, 0.07–0.08; P < 0.001) (Fig. 2A). Using multivariate analysis to adjust for other co-variables, we consistently observed that elevated levels of Kyn (odds ratio [OR], 3.10; 95 % CI, 1.92–5.00; P < 0.001) and the K:T ratio (OR, 3.96; 95 % CI, 1.89–8.30; P < 0.001) were significant risk factors for Bramet in our patient cohorts (Table 3).

Fig. 2.

Metabolites of IDO1 and mRNA expression are elevated NSCLC patients with Bramet. (A) IDO1 metabolites in all patients (upper panel) or patients with distant metastasis (lower panel), categorized as those with (Yes) and without (No) brain metastasis, are shown. (B). Kaplan-Meier curves showing a significant correlation between the baseline levels of Kyn and survival in Bramet patients. (C) Kaplan-Meier curves display the relationships between the baseline levels of K:T ratio and survival in Bramet patients. Log-rank test was used. The color shaded area represents the 95 % confidence intervals. (D) Violin plots show the IDO1 expression [Log₂(CPM+1)] in primary and metastatic tissues. (E) Violin plots show the AUCell score from Trp metabolic pathway in primary and metastatic tissues. (F) Heatmap demonstrates associations between IDO1 metabolites and IDO1 pathway genes. Pearson correlation analysis was used. *, P < 0.05. AADAC, arylacetamide deacetylase; AUC, area under curve; Bramet, brain metastasis; DDC, diethyldithiocarbamate; IDO1, Indoleamine 2,3-dioxygenase; Kyn: kynurenine; K:T ratio, kynurenine vs. tryptophan; mLN, metastatic lymph nodes; nLN, normal lymph nodes; Pre: blood collected before tumor resection; Pre/ Post-K:T ratio: relative K:T ratio tested after and before radiotherapy; Post, blood collected after tumor resection; tLung, primary lung tumor; TPH, tryptophan hydroxylase; Trp, tryptophan.

Table 3.

Univariate analysis and multivariate analysis of patient characteristics and IDO1 metabolites for the risk of brain metastasis in all patients.

Variable	Univariate OR (95 % CI)	P-value	Multivariate OR (95 % CI)	P-value
Smoking status
Non-smoking	Reference	Reference
Chronic smoker	0.73 (0.32–1.68)	0.461
Ex-smoker	0.85 (0.32–2.25)	0.748
Cancer subtype
ADC	Reference	Reference
SCC	0.35 (0.12–1.07)	0.066	0.49 (0.12–2.07)	0.334
Others	5.56 (1.37–22.52)	0.016	1.74 (0.22–13.63)	0.598
EGFR mutant
No	Reference	Reference
Yes	2.02 (0.87–4.70)	0.102
PD-L1 expression, %	0.99 (0.97–1.02)	0.466
Previous treatment
No	Reference	Reference
Chemotherapy	46.67 (6.02–361.55)	< 0.001	17.60 (2.11–146.58)	0.008
Targeted therapy	51.13 (6.49–402.77)	< 0.001	25.10 (3.02–208.60)	0.003
Others	16.00 (1.82–140.92)	0.012	13.43 (1.44–125.70)	0.023
Age, years	0.99 (0.96–1.02)	0.470
Gender
Male	Reference	Reference
Female	0.88 (0.46–1.68)	0.701
Kyn, µM	3.45 (2.22–5.35)	< 0.001	3.10 (1.92–5.00)	< 0.001
Trp, µM	0.93 (0.89–0.96)	< 0.001	0.92 (0.89–0.96)	< 0.001
K:T ration, ln	4.96 (2.44–10.05)	< 0.001	3.96 (1.89–8.30)	< 0.001

The bold values mean the differences between these comparing groups are statistical significance, which P value < 0.05.

Abbreviations: ADC, adenocarcinoma; CI, confidence intervals; IDO1, indoleamine 2,3-dioxygenase; K:T ratio, Kynurenine vs. Tryptophan; Kyn, Kynurenine; OR, odds ratio; SCC, squamous cell carcinoma; Trp, Tryptophan.

Among patients with distant metastasis, Kyn levels were higher in Bramet patients (n = 54; 2.7 µM; 95 % CI, 2.4–3.0 µM) compared to extra-cranial metastases (n = 45; 1.9 µM, 95 % CI, 1.5–2.2 µM; P < 0.001). The K:T ratio was also higher in Bramet patients (0.12; 95 % CI, 0.10–0.15) compared to extra-cranial metastases (0.10; 95 % CI, 0.08–0.11; P = 0.028). Trp levels remained unchanged (Fig. 2A and Supplemental Fig. 3A). Multivariate analysis further demonstrated that the levels of Kyn (OR, 3.71; 95 % CI, 2.04–6.75; P < 0.001) and the K:T ratio (OR, 6.49; 95 % CI, 2.43–17.33; P < 0.001) were risk factors for Bramet among the patients with distant metastasis (Table 4).

Table 4.

Univariate and multivariate analysis of patient characteristics and IDO1 metabolites for the risk of brain metastasis in patients with distant metastasis.

Variable	Univariate OR (95 % CI)	P-value	Multivariate OR (95 % CI)	P-value
Smoking status
Non-smoking	Reference	Reference
Chronic smoker	0.52 (0.19–1.41)	0.198
Ex-smoker	0.68 (0.21–2.22)	0.519
Cancer subtype
ADC	Reference	Reference
SCC	0.41 (0.12–1.45)	0.167
Others	3.25 (0.64–16.60)	0.157
EGFR mutant
No	Reference	Reference
Yes	2.44 (0.93–6.42)	0.071
PD-L1 expression, %	1.00 (0.97–1.03)	0.937
Previous Treatment
No	Reference	Reference
Chemotherapy	10.00 (1.15–86.75)	0.037	5.55 (0.53–57.65)	0.152
Targeted therapy	9.88 (1.12–86.98)	0.039	6.64 (0.64–68.87)	0.113
Others	9.60 (0.88–105.17)	0.640	16.12 (1.20–217.33)	0.066
Age, years	0.98 (0.94–1.02)	0.252
Gender
Male	Reference	Reference
Female	1.66 (0.73–3.74)	0.225
Kyn, µM	3.15 (1.83–5.43)	< 0.001	3.71 (2.04–6.75)	< 0.001
Trp, µM	0.94 (0.90–0.98)	0.001	0.93 (0.89–0.97)	0.060
K: T ratio, ln	5.83 (2.34–14.54)	< 0.001	6.49 (2.43–17.33)	< 0.001

The bold values mean the differences between these comparing groups are statistical significance, which P value < 0.05.

Abbreviations: ADC, adenocarcinoma; CI, confidence intervals; IDO1, indoleamine 2,3-dioxygenase; K: T ratio, Kynurenine vs. Tryptophan; Kyn, Kynurenine; OR, odds ratio; SCC, squamous cell carcinoma; Trp, Tryptophan.

Among patients with Bramet, those with elevated levels of Kyn exhibited worse survival outcomes (median OS, 6.9 months) in comparison to patients with lower Kyn concentrations (median OS, 11.5 months; P = 0.011; cut-off point = 3.14) (Fig. 2B). Similarly, patients with Bramet who had high K:T ratios (median OS, 8.0 months) experienced worse survival compared to those with a low K:T ratio (median OS, 11.8 months; cut-off point = 0.119; P = 0.039) (Fig. 2C). Additionally, multivariate analysis demonstrated that elevated levels of Kyn (OR, 1.76; 95 % CI, 1.29–2.41; P < 0.001) and a higher K:T ratio (HR, 2.30; 95 % CI, 1.40–3.77; P = 0.001) were identified as significant risk factors for OS. Other factors were not associated with OS in this subgroup (Table 5).

Table 5.

Univariate and multivariate Cox regression analysis of patient characteristics and IDO1 metabolites for overall survival among brain metastasis patients.

Variable	Univariate HR (95 % CI)	P-value
Smoking
Non-smoking	Reference	Reference
Chronic smoker	0.75 (0.35–1.62)	0.470
Ex-smoker	0.80 (0.33–1.95)	0.622
Subtype
ADC	Reference	Reference
SCC	0.82 (0.35–1.94)	0.651
Other	0.83 (0.30–2.35)	0.731
EGFR mutant
No (n = 17)a	Reference	Reference
Yes (n = 18)a	0.87 (0.45–1.68)	0.680
PD-L1 expression, %	0.99 (0.97–1.01)	0.199
Previous treatment
No	Reference	Reference
Chemotherapy	2.07 (0.49–8.83)	0.326
Targeted therapy	1.71 (0.40–7.38)	0.473
Others	0.95 (0.17–5.23)	0.952
Age, years	0.99 (0.96–1.02)	0.515
Gender
Male	Reference	Reference
Female	1.22 (0.68–2.18)	0.501
Kyn, µM	1.76 (1.29–2.41)	< 0.001
Trp, µM	0.99 (0.96–1.02)	0.526
K:T ratio, ln	2.30 (1.4–3.77)	0.001
Post-treatment
Surgery	Reference	Reference
Radiotherapy	2.27 (0.93–5.53)	0.072
Chemotherapy	1.33 (0.16–11.29)	0.792
Targeted therapy	1.22 (0.14–10.34)	0.855
Immunotherapy	3.73 (0.43–32.11)	0.230

The bold values mean the differences between these comparing groups are statistical significance, which P value < 0.05.

^aOther patients with no genomic detection.

Abbreviations: ADC, adenocarcinoma; CI, confidence intervals; HR, hazard ratio; IDO1, indoleamine 2,3-dioxygenase; K:T ratio, Kynurenine vs. Tryptophan; Kyn, Kynurenine; SCC, squamous cell carcinoma; Trp, Tryptophan.

To further analyze the IDO1 mRNA expression at the cell level, a total of 239,787 cells from single-cell RNA analysis from different tissues including normal lymph nodes (nLN, n = 10), primary lung tumor (tLung, n = 24), metastatic lymph nodes (mLN, n = 7), Bramet (mBrain, n = 15), and extra-Bramet (extra mBrain, n = 1) were performed in patients with NSCLC, respectively. Cells were clustered and plotted in the t-SNE maps split by the different tumor types (Supplemental Fig. 3B). The IDO1 expression was significantly higher in the mBrain sites (mean, 0.26) and mLN tissue (0.20), compared with metastatic lesions of other tissue sites: tLung (0.13), nLN (0.04) and extra-cranial metastasis (0.07; P < 0.001) (Fig. 2D). Furthermore, the AUCell algorithm was used to score each cell for enrichment of specific transcriptomes in Trp metabolic pathway. We found that the AUCell score from Trp metabolic gene signature was increased in patients’ samples from stage IV (Fig. 2E).

To investigate the potential association between tumor tissue IDO1 mRNA expression and circulating IDO1 metabolites, Bramet tissue samples and blood samples from the same eighteen Bramet patient were tested with IDO1 pathway genes mRNA expression and IDO1 metabolites, respectively. It was found that the mRNA expression level of IDO1 was significantly correlated with the baseline Kyn concentrations in the circulations of these Bramet patients (r = 0.50, P = 0.03) (Fig. 2F).

3.4. mRNA expression pattern of IDO1 and its potential regulatory mechanisms in Bramet tissue

The differential mRNA expression pattern of IDO1 in distinct cell types within different tumor tissues was further analyzed. The epithelial cells (34 %) were predominant in the lung primary tumor. B cells (26 %) and CD4⁺ T (58 %) cells were enriched in the nLN tissues. Tumor tissues from mLN were mainly composed of CD4⁺ T cells (22 %), B cells (12 %), and epithelial cells (32 %). The extracranial metastatic tumor tissues were mainly composed of epithelial cells (73 %). CD4⁺ T cells (20 %) and epithelial cells (42 %) were the two major cell types in the Bramet tumors (Fig. 3A). In the Bramet lesions, IDO1 mRNA was more abundantly expressed in endothelial cells (0.43), epithelial cells (0.40), and mesenchymal cells (0.31) than those in other cell types: B cells (0.15), CD4⁺ T cells (0.11), CD8⁺ T cells (0.11), mast cells (0.07) and myeloid cells (0.18; P < 0.005). Besides, IDO1 mRNA was abundantly expressed in B cells (0.2) compared with other cell types: endothelial cells (0.07), epithelial cells (0.06), and mesenchymal cells (0.09), CD4⁺ T cells (0.08), CD8⁺ T cells (0.12) and myeloid cells (0.07) within the extra-cranial metastasis site (P < 0.005). In the mLN tissues, the expression of IDO1 was evenly expressed in the various cell types: CD4⁺ T cells (0.22), CD8⁺ T cells (0.18), epithelial cells (0.18), mast cells (0.25), B cells (0.23), and mesenchymal cells (0.21). In the nLN tissue, IDO1 expression was significantly more abundant in endothelial cells (0.48), mast cells (0.33), and myeloid cells (1.00), than those in other cell types (P < 0.005). In the primary lung tumor tissue, IDO1 was significantly more abundant in endothelial cells (0.17), epithelial cells (0.15), and myeloid cells (2.50) than other cell types CD4⁺ T cells (0.08), CD8⁺ T cells (0.04), B cells (0.10), mesenchymal cells (0.07) and mast cells (0.05; P < 0.005) (Fig. 3B).

Fig. 3.

mRNA expression pattern of IDO1 and its potential regulatory mechanisms in Bramet tissue. (A) The percentages of each cell type are demonstrated in a bar graph. (B) Violin plot shows the IDO1 mRNA expression [Log₂(CPM+1)] in distinct cell types from primary and metastatic tissues. (C) tSNE plot shows the spatial RNA-seq barcoded spots separated by distinct cell clusters with different colors. (D) UMAP plot show the IDO1 spatial mRNA expression pattern in the brain metastasis tissues from lung cancer. (E) IDO1 and featured proteins interaction network in Bramet. (F) Violin plots show the AUCell score from IDO1 regulating genes in primary and metastatic tissues. Data are analyzed by Post-Hoc tests (Bonferroni correction) and displayed as mean ± standard errors. *, ^⁎⁎, ^⁎⁎⁎, and^⁎⁎⁎⁎ indicates P < 0.05, P < 0.01, P < 0.005, and P < 0.001, respectively. Extra mBrain: Extra brain metastasis; mBrain, brain metastasis; mLN, metastatic lymph nodes; NK, naturally killer cells; nLN, normal lymph nodes; tLung, primary lung tumor.

To further investigate the spatial mRNA expression pattern of IDO1 in Bramet patients, Bramet tissue (metastasized to the sellar region, n = 1) from patients with primary NSCLC was used for single-cell spatial transcriptomics sequencing. A total of 1008 individual spots were obtained, and eight cell clusters were defined (Fig. 3C). Furthermore, IDO1 was highly expressed in the cluster 1, which was featured as macrophage markers: CD70⁺S100P⁺ (P < 0.001; Fig. 3D). To identify the potential proteins that may interact with IDO1 in the cluster 1, the protein-protein interaction analysis was performed using the cell cluster 1-specific genes (P < 0.05). Chemokines such as CXCL11, CCL8 and CXCR3 directly interacted with IDO1 in the cluster 1 (Fig. 3E). Furthermore, the AUCell score from the genes potentially modulating of IDO1 pathway was enriched in myeloid cells from scRNA-seq data integrated analysis (Fig. 3F).

3.5. Metabolites of IDO1 effect on the functions of immune cells

We further investigated the potential link between the immune cell profile and IDO1 metabolites in vivo immune cell subtyping (Supplemental Fig. 2) and in vitro culture treatment. There was a weak negative correlation between the Kyn concentration and the percentage of immune cells, including total lymphocytes (r = −0.32, P = 0.08), and CD4⁺ T cells (r = −0.36, P = 0.048). Conversely, the percentage of immune cells such as B cells (r = 0.25, P = 0.19), monocytes (r = 0.06, P = 0.77), total T cells (r = −0.07, P = 0.72), and CD8⁺ T cells showed no significant association with Kyn concentration in patients. Notably, the percentage of Tregs exhibited a positive correlation with Kyn concentration in patients with Bramet (r = 0.65, P < 0.001). Furthermore, the association between the K:T ratio and the percentages of immune cells was analyzed. The percentage of total lymphocytes showed a borderline significant correlation with the K:T ratio (r = - 0.36, P = 0.051). Meanwhile, the percentage of B cells (r = 0.22, P = 0.23), monocytes (r = −0.16, P = 0.40), total T cells (r = −0.05, P = 0.77), CD8⁺ T cells, NKT cells and CD4⁺ T cells (r = −0.22, P = 0.25) displayed no significant correlations with the K:T ratio. The percentage of Treg cells was positively associated with the K:T ratio in patients with Bramet (r = 0.67, P < 0.001) (Fig. 4A).

Fig. 4.

Metabolites of IDO1 on the functions of immune cells. (A) Scatter plots showing the Pearson correlations between the concentration of Kyn or K:T ratio and the percentages of immune cell subtypes from brain metastasis patients. The gray shaded area indicates the 95 % confidence intervals. (B) Representative flow cytometry plots of the expression of T cell exhaustion markers. (C-E) Quantitation of T cell dysfunction markers, including PD-1 (C), LAG3 (D) and TIM3 (E). Data are analyzed by One-way ANOVA test and displayed as mean ± standard errors. *, ^⁎⁎, ^⁎⁎⁎, and^⁎⁎⁎⁎ indicates P < 0.05, P < 0.01, P < 0.005, and P < 0.001, respectively. Bramet, brain metastasis; Kyn, kynurenine; NKT, naturally killer cells; Treg, regulatory T cells.

For in vitro validation, total T cells were treated with Kyn in different concentrations (Supplemental Fig. 5) or serum—obtained rom patients with Bramet or from stage IV patients (Fig. 4B). T cells treated with serum from Bramet patients exhibited higher levels of dysfunctional markers (PD-1, LAG3 and Tim-3), which was consistent with Kyn treatment (Fig. 4C-E).

3.6. Identification of other plasma metabolites for brain metastasis

Using the untargeted metabolomics, partial least squares-discriminant (PLS-DA) analysis was used to investigate the differences between the stage IV patients with (n = 13) or without Bramet (n = 12) (Supplemental Fig. 6). PLS-DA plot showed a clear separation of the different metabolic profiles using the first two principal components (PCs) with 11.6 % and 10.8 % of explained variance in the ESI (+) model and PCs with 6.7 % and 6.1 % in the ESI (-) model (Fig. 5A). The top 15 metabolites contributing to variation from PLS-DA in the metabolic differences between patients with or without Bramet are shown in variable importance plots (VIP) plots (Fig. 5B and C).

Fig. 5.

Metabolic profile difference between patients with or without Bramet. (A) Partial least squares discriminant analysis (PLS-DA) plots show the clear metabolic separation between the patients with (Yes) or without Bramet (No) in the ESI (+) and ESI (-) model, respectively. (B, C) Variable importance of projection plot for the vital metabolites significantly contribute to the difference between the patients with (Yes) or without Bramet (No) in the ESI (+) (B) and ESI (-) (C) models, respectively. (D) KEGG pathway functional enrichment analysis of the vital metabolites. ESI, electrospray ionization.

Besides, a total of 183 variable metabolites (VIP > 1) by matching the in-house and public databases. Functional analyses of these differential metabolites were predominantly enriched in the metabolic pathway terms such as “phospholipid biosynthesis”, “phosphatidylcholine biosynthesis” or “beta oxidation of very long chain fatty acids”, as well as the “tryptophan pathway”, the IDO1 associated pathway (Fig. 5D).

4. Discussions

In this study, we identified the potential of IDO1 metabolites as biomarkers for NSCLC patients with Bramet, and its role in promoting the immune-suppressive cell functions and negatively impact patient outcomes. This study demonstrated that 1) K:T ratio was significantly increased in patients with NSCLC and patients with stage IV disease had significantly higher levels of Kyn and K:T ratio at baseline; 2) patients with Bramet had significantly higher level of IDO1 activity, i.e., high level of Kyn or high K:T ratio, and worse survival; 3) other plasma metabolites such as phospholipid pathway was also detected; and 4) the high level of Kyn in serum from Bramet patients was significantly associated impaired immune status indicated as higher percentage of immune suppressive cells, elevated expression of T cell dysfunctional markers.

Patients with NSCLC had higher IDO1 activity levels than normal controls. In addition, the Trp concentration in plasma was significantly lower in lung cancer patients than in the normal controls. These findings were consistent with previous publications, demonstrating that patients with lung cancer had significantly lower Trp concentrations (i.e. high IDO1 activity) in circulation than healthy controls.^26,27 This difference may be caused by the low degradation rate of Trp in normal conditions, but can be significantly enhanced by proinflammatory stimuli such as interferon-gamma (IFN-γ) from the TIME.²⁸ This result suggests the potential role of IDO1-associated metabolite Trp as a tumor marker for NSCLC.

Moreover, patients with advanced-stage NSCLC had significantly higher IDO1 activity levels than those in the earlier stages of diseases. These findings were also consistent with the meaningful findings from Suzuki Y. et al., who demonstrated that the plasma K:T ratio was increased in advanced-stage lung cancer patients.²⁷ These results were consistent with the clinical findings in the blood IDO1 measurement from this study. Another one published study that showed IDO1 expression was increased in advanced-stage cancer.²⁹ However, in a study of 28 patients with lung cancer, no significant association between the tissue IDO1 mRNA expression and cancer disease staging was observed.³⁰ The inconsistencies in these findings may be attributed in part to variations in sample sizes and approaches used to measure IDO1 expression, as well as tumor heterogeneity. It was also important to note that both studies had relatively small sample sizes and included heterogeneous patient populations. Despite these discrepancies, these data highlight the significant role of heightened IDO1 activity in STIE during lung cancer progression. Larger studies with more uniform treatments are necessary to validate the potential role of IDO1 metabolities as an indicator for advanced stages of NSCLC.

It is interesting to note that circulating IDO1 metabolites were significantly higher in patients with Bramet. This has not been reported previously, but it makes biological sense, as extra aggressiveness is needed for tumors to penetrate through blood-brain barrier (BBB) and into the brain. It was consistent with the findings that IDO1 mRNA expression was abundantly expressed in brain and metastatic tumors.^31,32 Consistently, patients with high expression of IDO1 had higher risk of developing distant metastasis.³³ It was also previously demonstrated that a low Kyn concentration was associated with a lower risk of distant metastasis and more prolonged survival (median OS, 41 month) even in metastasis patients.^23,25 Mechanistically, IDO1 expression in tumor cells can have positive association with the tumor growth and tumor sizes in mouse models.³⁴ We observed that the expression of IDO1 in stage I tumors was notably higher compared to stages II and III. This pattern may suggest an initial immune response (immune equilibrium) or reflect unique tumor microenvironmental factors specific to stage I tumors.³⁵ On the other hand, higher IDO1 activity can help the tumor cell to induce the immune suppressive microenvironment to evade the immune surveillance.³⁶ Indeed, the single-cell analysis found that the IDO1 was highly expressed in the cancer epithelial and mesenchymal cells in Bramet tissues, compared with other tissue origins. These results also reveal the potential roles of host immunity mediated by IDO1 in modulating Bramet progression in NSCLC patients.

This study, for the first time to our knowledge, showed that the higher Kyn concentrations or K:T ratios were related to worse OS and the circulating immune suppressive functions in patients with Bramet. This finding was consistent other cancer-related studies regarding with the crucial roles of IDO1-mediated immune suppression through modulating the systemic immune response. For example, injection of an enzyme (kynureninase) that can degraded the systemic level of Kyn was associated with a remarkably increase in the proliferation of and tumor and infiltration of polyfunctional CD8⁺ T cells.³⁷ In another study, the peripheral Treg/CD8⁺ cytolytic T cell ratio was also increased when the IDO1 expression was elevated in brain, which could further increase the incidence of primary brain tumor and mortality rate.³⁸ It was also reported that the elevated human brain IDO1 mRNA expression was associated with the increased percentages of Treg cells and the decrease percentages of CD8⁺ cytolytic T cells in the peripheral blood from aged blood samples.³⁹ These results further revealed the simultaneous immune suppressive functions of IDO1 in both TIME and STIE.

It was noteworthy that Kyn can modulate the expression of dysfunctional markers on T cells, consistent with previous findings indicating that Kyn derived from tumor cells induces CD8⁺ T cell dysfunction within the tumor microenvironment via the aryl hydrocarbon receptor (AHR) pathway.^40,41 It would be valuable to conduct additional experiments to investigate the prognostic significance of Kyn and dysfunctional T cell markers in patients with Bramet. Additionally, the observed high mRNA expression of IDO1 coupled with low CD8⁺ T cell infiltration underscores the pivotal immune-suppressive regulatory role of IDO1 in the progression of Bramet within the tumor microenvironment. Furthermore, the analysis of melanoma Bramet samples revealed high expression of IDO1 in microglia.³¹ Single-cell RNA analysis of IDO1 expression suggested that IDO1 was up-regulated in the non-T cell types in Bramet regions, such as macrophage cells, which were associated with poor survival.⁴¹ Furthermore, a high mRNA expression level of IDO1 was always associated with low infiltration of CD8⁺ T cells, which also suggested the master immune-suppressive regulatory roles of IDO1 in Bramet progression in the brain TIME.⁴² Consistently, IDO1 mainly was also found to be highly expressed in the microglia in the analysis of melanoma Bramet samples.³¹ This study further revealed the relationship between heterogenous IDO1 expression and the tumor progression within the tumor microenvironment. However, more experiments shall be conducted to confirm these findings.

IDO1 inhibitors treatment in combination with ICIs showed promising results in a pre-clinical study, yet failed to evince improved outcomes in clinical trials. Epacadostat was one of the most advanced IDO1 inhibitors, reaching phase 3 clinical trials in combination with pembrolizumab for various cancers. However, it failed to meet its primary endpoint in melanoma, leading to the termination of several trials.⁴³ The possible reasons why these trials failed may be that the expression and regulation of IDO1 are context-dependent in both TIME and STIE.^44,45 Indeed, our study revealed the IDO1 heterogeneous expression patterns in STIE during cancer progression and treatments. These findings suggest the need to identify an optimal window of opportunity for using IDO1 inhibitors in cancer treatment. Furthermore, the study underscores the importance of completely depleting the circulating Kyn pool, rather than relying solely on single-use IDO1 inhibitors, which only partially reduce systemic Kyn levels.

Our untargeted metabolomics analysis also suggested the association IDO1 metabolites interacting with other metabolic pathways in Bramet. Recent studies also suggested that overproduction of reactive oxygen species resulting from IDO1 activity and downstream metabolites like quinolinic acid could potentially led to the destruction of phospholipids in cell membranes.⁴⁶ Other studies suggest that IDO1 and its metabolites interact with ferroptosis suppression, 1-carbon metabolism, and NAD⁺ metabolism, indicating broader metabolic roles beyond immunomodulation.^47,48 Besides, the metabolism of tryptophan by IDO1 contributes one-carbon units to the tetrahydrofolate cycle, which is linked to the tricarboxylic acid (TCA) cycle for the biosynthesis of nucleotides and other biomolecules.⁴⁹ Restricting serine and glycine enhanced the anti-tumor activity of the IDO1 inhibitor epacadostat in a pancreatic cancer mouse model, suggesting that tryptophan metabolism via IDO1 provides an alternative one-carbon source when serine is limited.⁵⁰ These findings suggest potential applications for Bramet treatment by combining IDO1 inhibitors with other metabolic inhibitors.

The study was limited in that the limited number of available samples for each stage, which may affect the interpretation of the statistical results, especially for intergroup comparisons. Data for this part need validation with large sample size validation. In addition, the impact of the previous therapies in altering the IDO1 activity was difficult to determine because it was common for cancer patients diagnosed with Bramet after multiline treatments. Future efforts to assemble the effects of therapies and clinical outcomes can be beneficial to broaden the application of IDO1 metabolites as biomarkers in Bramet.

5. Conclusions

In summary, this study has validated the pioneering significance of IDO1 metabolites as innovative biomarkers in the realm of NSCLC Bramet. This revelation not only extends knowledge of immune-suppressive mechanisms to promote Bramet but also encourages us to adopt a paradigm-shifting perspective in our approach to cancer diagnosis and treatment.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.