Skip to main content
NCBI home page
Annals of Laboratory Medicine logoLink to Annals of Laboratory Medicine
. 2026 Apr 2;46(5):539–553. doi: 10.3343/alm.2025.0536

Identification of Metabolic Biomarkers in Bone Marrow Aspirate for Diagnosis, Prognosis, and Therapeutic Monitoring in Pediatric Langerhans Cell Histiocytosis

Xiaodan Liu 1,2,3,*, Jiayi Wang 2,*, Yali Wang 4, Huilin Lin 2, Qingqing Zheng 2, Mansi Cai 2, Xinxin Cao 5, Hua Jiang 2, Xinliang Mao 3,✉
PMCID: PMC13458183  PMID: 41923546

Abstract

Background

Metabolic reprogramming is frequently observed but poorly understood in Langerhans cell histiocytosis (LCH), a rare inflammatory myeloid neoplasm. To improve molecular classification, treatment monitoring, and prognosis, we comparatively analyzed metabolite alterations in bone marrow aspirates (BMAs) from patients with various subtypes of LCH and healthy individuals.

Methods

Metabolite alterations in BMAs were identified using untargeted metabolomic profiling and quantified using targeted metabolomics. ROC curve and Kyoto Encyclopedia of Genes and Genomes analyses were used to identify potential metabolic biomarkers for diagnosis, prognosis, and therapeutic monitoring in pediatric LCH.

Results

The major metabolites altered in LCH included lipid/lipid-like molecules and organic acids and their derivatives, primarily involved in pathways related to choline metabolism, glycerophospholipid metabolism, and unsaturated fatty acid biosynthesis. Nine metabolites were identified as potential diagnostic biomarkers. 1-Heptadecanoyl-sn-glycero-3-phosphocholine (PC[170/00]) levels varied significantly among LCH clinical subtypes. L-kynurenine and betaine aldehyde levels helped in effectively distinguishing inflammatory from cancer-like LCH, and increased L-kynurenine and 2-arachidonoyl-1-stearoyl-sn-glycero-3-phosphoethanolamine levels and reduced PC(170/00) levels predicted poor prognosis. Several metabolites, including dodecanoic acid and inosine, may serve as indicators for treatment efficacy assessment in LCH with multisystem and risk organ involvement. L-kynurenine and betaine aldehyde levels were verified using targeted quantitative metabolomics, and their clinical significance was assessed.

Conclusions

Alterations in BMA metabolic profiles within the LCH population and in specific metabolites may provide valuable insights into LCH pathogenesis and serve as biomarkers for diagnosis, prognosis, and therapeutic response assessment.

Keywords: Biomarker, Bone marrow, Kynurenine, Langerhans cell histiocytosis, Metabolomics

INTRODUCTION

Langerhans cell histiocytosis (LCH), a rare inflammatory myeloid neoplasm characterized by the accumulation of pathological CD1a+ and CD207+ cells in lesions, presents with highly variable clinical manifestations [1]. LCH can affect patients of all ages but predominantly occurs in children. Diagnosis relies on CD1a and CD207 staining of lesion tissues, and disease severity is clinically assessed based on lesion extent, ranging from isolated, indolent, to explosive multisystem involvement [2]. Children with LCH involving the liver, spleen, or bone marrow have a mortality risk of approximately 15%. Despite treatment advances, reactivation rates remain >30%, often leading to serious long-term neurological or endocrine complications that significantly impair quality of life [2]. The molecular characteristics of LCH have been partially elucidated and include BRAF variants (e.g., BRAFV600E) and MAP2K1 and ARAF variants, which drive constitutive mitogen-activated protein kinase (MAPK) pathway activation [3, 4]. Currently, treatment decisions rely solely on clinical presentation because reliable molecular markers for risk stratification are lacking.

Metabolic reprogramming is a fundamental characteristic of cancer cells and plays a crucial role in abnormal proliferation and carcinogenesis [5]. Metabolomic analysis is a powerful tool for monitoring disease status and unraveling the pathogenesis of various diseases [6]. Reprogrammed metabolic pathways have been widely investigated in the contexts of cancer development, progression, and treatment resistance [7, 8]. Advances in high-throughput metabolomics technologies have enabled comprehensive identification and quantification of endogenous low-molecular-weight metabolites, which reflect complex interplays among host factors, genetic influences, and environmental exposures [9]. The MAPK pathway, which is constitutively activated in LCH, is a master regulator of cellular metabolism, influencing glucose uptake, aerobic glycolysis, and biosynthetic processes essential for cell growth and proliferation [10, 11]. BRAF, a driver gene in LCH, has been implicated in metabolic reprogramming in numerous disorders [12, 13]. However, metabolic changes in LCH pathophysiology and treatment remain unclear.

To address the above gaps, we profiled global metabolism in bone marrow aspirates (BMAs) from 34 patients with LCH of three subtypes: single-system disease (SS), multisystem disease without risk organ involvement (MS-RO–), and multisystem disease with risk organ involvement (MS-RO+). We mapped the metabolomic profiles of these subtypes and explored metabolic changes associated with treatment outcomes in high-risk LCH to improve molecular classification, treatment monitoring, and prognosis in affected children.

MATERIALS AND METHODS

Detailed methods are described in the Supplemental Materials.

Patient population and clinical data collection

In total, 34 patients diagnosed as having LCH, including 28 newly diagnosed (ND) cases and six with complete remission (CR), were enrolled in this study conducted at the Guangzhou Women and Children’s Medical Center of Guangzhou Medical University between January 2017 and June 2023 (Supplemental Data Table S1). Patients were followed up until December 31, 2023. Control groups matched for age and sex, comprising 10 individuals for untargeted and 12 additional for targeted metabolomics analyses, were established during the same period. All patients, except for one who succumbed to rapid disease progression, received systemic therapy. Demographic characteristics and clinical diagnostic information are collected from the medical record system. Particularly, we categorized LCH into two groups based on clinical symptoms [14]. Rash, otorrhea, bone pain, eosinophilic granuloma of the long bones, and diarrhea (characterized by an eosinophilic granulomatous inflammatory background) were classified as inflammatory symptoms (IS), whereas non-infectious persistent fever, gingival or palatal swelling, and thrombocytopenia (characterized by invasive malignant features) were classified as tumor-like symptoms (TLS). On the follow-up date, patients were classified into a good prognosis group (GPG) if they achieved CR, or a poor prognosis group (PPG) if they experienced recurrence or death.

Liquid chromatography mass spectrometry (LC-MS)–based metabolomics and data QC

For untargeted metabolomics of polar metabolites, extracts were analyzed using a quadrupole time-of-flight mass spectrometer (AB6600, Framingham, MA, USA) coupled to a hydrophilic interaction chromatography system in electrospray ionization mode at Shanghai Applied ProteinTechnology (Shanghai, China), as described previously [15]. The raw MS data (wiff.scan files) were converted to mzXML files using ProteoWizard MSConvert and imported into XCMS software (https://xcmsonline.scripps.edu/. Only variables having >50% non-zero measurement values in at least one group were maintained. Metabolites were identified from MS/MS spectra using an in-house database established with available authentic standards. We used the Metabolomics Standards Initiative level of confidence to grade metabolite annotation confidence (levels 1–4) [16].

Targeted metabolite quantitation

Betaine aldehyde and L-kynurenine standards (Yuanye, Shanghai, China) were quantified using deuterated L-leucine-d3 (Macklin, Shanghai, China) as internal standards using an LC-MS/MS system (Thermo Q Exactive Focus Orbitrap; Thermo Fisher, Waltham, MA, USA) in positive electrospray ionization and full-scan (m/z 100–1,000) modes. Separation was achieved on a C18 column with a water/acetonitrile gradient (detailed information is provided Supplemental Data Table S2). Metabolites in BMA from 34 patients and 22 healthy controls were quantified based on multipoint calibration curves (0–450 μM), using the isotope dilution method.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 8.0. Normally distributed data are presented as mean±SD, and non-normally distributed data are presented as median and interquartile range. Categorical variables were compared using Fisher’s exact test, and continuous variables were analyzed using the independent Student’s t-test or Mann–Whitney U test. Multiple groups were compared using the Kruskal–Wallis test with Bonferroni correction. ROC curves were used to calculate the area under the curve (AUC). Event-free survival (EFS), defined as the time from randomization to the first occurrence of disease progression, relapse, or death from any cause, was analyzed using the Kaplan–Meier method, and outcomes were compared among subgroups using the log-rank test. Significance was set to P<0.05.

Ethics

The study was approved by the Ethics Committee of Guangzhou Women and Children’s Medical Center (Approval No. [2024]024A01) and conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from the patient’s parents or guardians.

RESULTS

Metabolic reprogramming of ND pediatric LCH

We profiled metabolites in BMAs from 28 patients with ND pediatric LCH and 10 healthy controls, using untargeted metabolomics analysis (Supplemental Data Table S3). The LC-MS data exhibited high reproducibility, with an average Spearman correlation coefficient of 0.90 in QC samples. In total, 1,516 metabolites were identified, including 895 and 621 detected in positive- and negative-ion modes, respectively (Fig. 1A). Lipids and lipid-like molecules (34.433%) and organic acids and derivatives (22.625%) were the two largest super classes (Fig. 1B). Volcano plots were generated to visualize differential metabolites identified through univariate statistical analysis for each mode, with fold change values >1.5 or <0.67 and P<0.05 (Fig. 1C and 1D). Orthogonal Partial Least Squares Discrimination Analysis(OPLS-DA) revealed distinct LCH and control clusters (Fig. 1E and 1F), with LCH metabolites clearly separated, indicating substantial BMA differences and model robustness, as validated using permutation analysis (Fig. 1G and 1H).

Fig. 1. OPLS-DA–based metabolomic profiling of pediatric patients with LCH and healthy controls. (A) Experimental design. BMA samples (N=34) were collected from pediatric patients with LCH at different treatment durations. Peripheral blood plasma samples from 10 healthy individuals were included as controls. (B) Proportion of metabolites identified in each chemical category. (C, D) Volcano plots for positive- (C) and negative-ion (D) modes. (E, G) OPLS-DA in positive- (E) and negative-ion (F) modes. Data points represent biological repeats in LCH (indicated in green) and control (indicated in blue) groups. t[1] indicates the score of samples on the first principal component to distinguish between groups, and to[1] indicates the score of samples on the second principal component to reflect intragroup variation. The ellipse represents the 95% confidence interval. (G, H) OPLS-DA permutation test in positive- (G) and negative-ion (H) modes.

Fig. 1

Open in a new tab

Abbreviations: LCH, Langerhans cell histiocytosis; OPLS-DA, orthogonal partial least squares discriminant analysis; Ctrl, control; ND, newly diagnosed; CR, complete remission; SS, single-system disease; MS-RO–, multisystem disease without risk organ; MS-RO+, multisystem disease with risk organ.

Based on a Variable Importance in the Projection (VIP) >1 and P<0.05, we identified 68 differential metabolites between patients with LCH and controls. Hierarchical clustering heatmaps revealed substantial differences in metabolite levels between the two groups, particularly in negative and positive ion modes (Fig. 2A). The top 10 differential metabolites are listed in Supplemental Data Table S4. To investigate the exact metabolic pathways reprogrammed in pediatric LCH, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis on the significantly altered metabolites, and 33 significantly affected metabolic and signaling pathways were identified (Supplemental Data Table S5). The bubble chart in Fig. 2B displays the top 10 significantly enriched signaling pathways. Choline metabolism, glycerophospholipid metabolism, and unsaturated fatty acid biosynthesis were the most prominent.

Fig. 2. Differential metabolites with potential value in the diagnosis of pediatric patients with LCH. (A) Hierarchical clustering heatmap of differential metabolites in positive- and negative-ion modes. Asterisks denote metabolites with the highest fold changes (>3 or <0.2). (B) Bubble diagram of KEGG pathway enrichment. The vertical axis represents the KEGG pathway, the horizontal axis represents the enrichment factor (i.e., the ratio of differential metabolite count to annotated metabolite count in a given pathway). (C–K) ROC curves for oleamide (C), docosatrienoic acid (D), Glu-Thr-Arg (E), myo-inositol (F), N-acetyl-D-glucosamine 6-phosphate (G), D-mannose (H), hypoxanthine (I), xanthine (J) and niacinamide (K) in patients with LCH. AUCs for distinguishing patients with LCH from healthy controls were all ≥90%.

Fig. 2

Open in a new tab

Abbreviations: KEGG, Kyoto Encyclopedia of Genes and Genomes; AUC, area under the ROC curve; PC(17:0/0:0), 1-heptadecanoyl-sn-glycero-3-phosphocholine; HDTAPC, 1-O-hexadecyl-2-deoxy-2-thio-s-acetyl-sn-glyceryl-3-phosphorylcholine; DPPC, 1,2-dihexadecanoyl-sn-glycero-3-phosphocholine; DSPS, 1,2-distearoyl-sn-glycero-3-phospho-l-serine; 18:1 lyso PG, 1-oleoyl-2-hydroxy-sn-glycero-3-phospho-(1’-rac-glycerol); SLPE, 1-stearoyl-2-linoleoyl-sn-glycero-3-phosphoethanolamine; APGPE, 2-arachidonoyl-1-palmitoyl-sn-glycero-3-phosphoethanolamine; MSPC, 1-stearoyl-2-hydroxy-sn-glycero-3-phosphocholine; 2-O-ETHYL-PAF, 1-O-hexadecyl-2-O-ethyl-sn-glycero-3-phosphocholine; CMPF, 3-carboxy-4-methyl-5-propyl-2-furanpropanoic acid.

Potential diagnostic metabolic biomarkers in pediatric LCH

To explore potential diagnostic biomarkers in pediatric LCH, we employed various feature selection methods and determined the weights of pre-screened metabolic markers. We selected 33 metabolites based on their weights (VIP >1 and P<0.01) (Fig. 2A) and analyzed their influence on sample classification using ROC curves. AUCs showed that top-ranking features such as fatty acyls [oleamide and docosatrienoic acid (DTA)], carboxylic acids and derivatives (Glu-Thr-Arg), organooxygen compounds (myo-inositol, N-acetyl-d-glucosamine 6-phosphate, and d-mannose), imidazopyrimidines (hypoxanthine and xanthine), and pyridines and derivatives (niacinamide) significantly enhanced the ability to distinguish patients with LCH from controls (Fig. 2C–2K). The ROC curves indicated that these diagnostic biomarkers exhibited excellent classification performance, with AUCs of >0.9. We further investigated these nine differential metabolites as potential diagnostic biomarkers because, except for oleamide and Glu-Thr-Arg, they all showed significantly increased abundances in LCH (P<0.001, Supplemental Data Fig. S1).

Metabolomics analysis of differential metabolites across distinct clinical subtypes

To investigate the role of metabolites in LCH classification, we divided the patients into SS and MS groups and compared their metabolites. Subsequently, 14 significantly differential metabolites were identified, 10 of which were lipid and lipid-like molecules (i.e., four glycerophospholipids, three fatty acyls, two prenol lipids, and one steroid derivative). Scatter plots for these metabolites are shown in Fig. 3A. Specifically, fatty acyl levels were markedly decreased, whereas glycerophospholipid levels were strikingly increased.

Fig. 3. Differences in relative abundances of differential metabolites among LCH subtypes. (A) Scatter diagram of differential metabolites between SS and MS LCH subtypes related to lipids and lipid-like molecules (*P<0.05, **P<0.01, ***P<0.001, Mann–Whitney U test). (B) Hierarchical clustering heatmap of differential metabolites among SS, MS-RO–, and MS-RO+ subtypes. (C) 1-Heptadecanoyl-sn-glycero-3-phosphocholine levels in SS, MS-RO–, and MS-RO+ groups (*P<0.05, **P<0.01, Kruskal–Wallis test and Bonferroni correction). (D) Hierarchical clustering heatmap of differential metabolites between patients with TLS and IS. (E) Scatter diagram of L-kynurenine and betaine aldehyde levels in the TLS and IS groups, and ROC curves showing their efficacy in distinguishing LCH subtypes. (*P<0.05, **P<0.01, ***P<0.001, Mann–Whitney U test).

Fig. 3

Open in a new tab

Abbreviations: SS, single-system disease; MS, multisystem disease; LPGPE, 2-linoleoyl-1-palmitoyl-sn-glycero-3-phosphoethanolamine; DHPDE, 1,2-dilinoleoyl-sn-glycero-3-phosphoethanolamine; OGPC, 1-oleoyl-sn-glycero-3-phosphocholine; PC(17:0/0:0),1-heptadecanoyl-sn-glycero-3-phosphocholine; 2-O-ETHYL-PAF, 1-O-hexadecyl-2-O-ethyl-sn-glycero-3-phosphocholine; PC O-36:3, 1-hexadecyl-2-(8z,11z,14z-eicosatrienoyl)-sn-glycero-3-phosphocholine; 1-N-M,[6-hydroxy-1-[2-(4-morpholinyl)ethyl]-1h-indol-3-yl]-1-naphthalenyl-methanone; TLS, tumor-like symptoms; IS, inflammatory symptoms.

Given that risk organs can be hidden and difficult to detect, we explored whether any metabolites in BMA could serve as predictors of multisystem risk organ involvement. We further classified patients into MS-RO– and MS-RO+ groups. Among six significantly differential metabolites between the two groups (Fig. 3B), 1-heptadecanoyl-sn-glycero-3-phosphocholine [PC(17:0/0:0)] was significantly reduced with increasing disease severity (Fig. 3C), showing promise as a potential biomarker for LCH subtyping.

Subsequently, we assessed whether BMA metabolites could distinguish between IS and TLS groups. In total, 87 metabolites were distinctly affected between the groups (the top 30 metabolites are listed in Supplemental Data Table S6). Hierarchical clustering heatmaps revealed substantial differences in the patterns of metabolite alterations between the TLS and IS groups (Fig. 3D); L-kynurenine and betaine aldehyde levels were significantly increased in patients with TLS and efficiently distinguished the two groups (Fig. 3E).

Roles of three metabolites in LCH prognosis prediction of children

The prognosis of pediatric LCH is reportedly strongly related to the BRAFV600E variant, multisystem involvement, risk organs, and age at diagnosis. To assess whether BMA metabolites can predict clinical outcomes in pediatric LCH, we performed a follow-up study (median, 657 days; range, 3–1,969 days). Among the 28 patients with ND LCH who were initially enrolled, three relapsed (10.7%) and two died (7.1%), resulting in an EFS rate of 82.1%. We compared the metabolite differences between the PPG and GPG groups (Fig. 4A). At the time of diagnosis, L-kynurenine and 2-arachidonoyl-1-stearoyl-snglycero-3-phosphoethanolamine (SAPE) levels were significantly elevated, whereas that of PC(17:0/0:0) was decreased in the subgroup of patients who relapsed or died (N=5) when compared with the levels in those who achieved continuous CR (Fig. 4B–4D).

Fig. 4. BMA metabolites are associated with the prognosis of pediatric LCH. (A) Hierarchical clustering heatmap of differential metabolites between GPG and PPG. (B–D) Scatter diagram of L-kynurenine, SAPE, and PC(17:0/0:0) levels in GPG and PPG (**P<0.01, Mann–Whitney U test). (E, G) Kaplan–Meier EFS curves for 28 pediatric patients with LCH stratified into high- and low-level groups based on median L-kynurenine, SAPE, and PC(17:0/0:0) levels (*P<0.05, **P<0.01, log-rank test, two-sided P values are indicated below the graphs).

Fig. 4

Open in a new tab

Abbreviations: GPG, good prognosis group; PPG, poor prognosis group; SAPE, 2-arachidonoyl-1-stearoyl-sn-glycero-3-phosphoethanolamine; PC(17:0/0:0), 1-heptadecanoyl-sn-glycero-3-phosphocholine; EFS, event-free survival.

Subsequently, the 28 patients were classified into low-level and high-level groups based on the median level of each metabolite, and their prognoses were compared. At the time of diagnosis, patients with high L-kynurenine and SAPE levels had a lower EFS rate than those with low levels, for whom the EFS rate was 100% around 2,000 days (Fig. 4E and 4F). An opposite pattern was observed for PC(17:0/0:0), for which lower levels were associated with poorer prognosis (Fig. 4G); these findings collectively suggest a potential association between these metabolites and prognosis.

Treatment-related metabolite alterations in patients with MS-RO+ LCH

To gain insight into metabolite alterations after treatment, we collected BMAs from six patients with MS-RO+ LCH following treatment-induced CR and compared them with those from ND patients (Fig. 1A). OPLS-DA revealed a distinct metabolite clustering pattern after treatment-induced CR in both positive and negative modes (Fig. 5A and 5B). The CR group was clearly separated from the ND group, indicating substantial metabolite differences and demonstrating the robustness of the analytical model, which was validated by permutation analysis (Fig. 5C and 5D).

Fig. 5. OPLS-DA-based metabolomic profiling of patients with ND and CR LCH and the clinical relevance of treatment-responsive metabolites. (A, B) OPLS-DA in positive- (A) and negative-ion (B) modes. Data points represent biological repeats in ND [pre-treatment (indicated in green)] and CR (indicated in blue) groups. (C, D) OPLS-DA permutation test in positive- (C) and negative-ion (D) modes. (E) KEGG pathway analysis of significantly dysregulated metabolites in ND and CR LCH, and the values indicate normalized z-score. (F) Hierarchical clustering heatmap of differential metabolites between ND and CR LCH. (G) Levels of 1,2-distearoyl-sn-glycero-3-phospho-l-serine, 1-oleoyl-2-hydroxy-sn-glycero-3-phospho-(1’-rac-glycerol), dodecanoic acid, and inosine detected in samples from patients with LCH at different disease stages. ***P<0.001; ns, not significant.

Fig. 5

Open in a new tab

Abbreviations: ND, newly diagnosed; CR, complete remission; OPLS-DA, orthogonal partial least squares discriminant analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; 18:1 Lyso PG, 1-oleoyl-2-hydroxy-sn-glycero-3-phospho-(1’-rac-glycerol); PE O-40:7, 1-(1z-octadecenyl)-2-(4z,7z,10z,13z,16z,19z-docosahexaenoyl)-sn-glycero-3-phosphoethanolamine; DSPS, 1,2-distearoyl-sn-glycero-3-phospho-l-serine.

Differential metabolite screening identified 69 metabolites. KEGG analysis revealed that these metabolites were predominantly enriched in pathways such as ABC transporters, steroid hormone biosynthesis, and primary bile acid biosynthesis, indicating their potential value in LCH therapeutic monitoring (Fig. 5E). Hierarchical clustering identified the top 20 differentially expressed metabolites (Fig. 5F), including four with significant post-treatment changes: 1,2-distearoyl-sn-glycero-3-phospho-l-serine (DSPS), 1-oleoyl-2-hydroxy-sn-glycero-3-phospho-(1’-rac-glycerol) (18:1 lyso PG), dodecanoic acid, and inosine. These metabolites were highly elevated in patients with ND MS-RO+ LCH but normalized after achieving CR (Fig. 5G).

Analysis of targeted metabolites by LC-MS/MS in patients with LCH

To validate their clinical relevance in pediatric LCH, we quantified BMA L-kynurenine and betaine aldehyde levels using LC-MS/MS. First, calibration curves were generated by spiking the authentic matrix with increasing analyte concentrations. Betaine aldehyde [m/z 138, retention time (RT) 1.55 min] and L-kynurenine (m/z 209, RT 2.15 min) levels were quantified based on multipoint calibration curves (0–450 μM) using L-leucine-d3 as an IS. Linear regression (1/x2 weighting, r2 >0.9) enabled absolute quantification. The calibration curves are shown in Supplemental Data Fig. S2. Next, we quantified L-kynurenine and betaine aldehyde levels in a cohort comprising 34 patients with LCH and 22 healthy controls. Levels of both metabolites tended to increase in patients with LCH compared with controls (Fig. 6A and 6B). Betaine aldehyde levels helped in effectively distinguishing patients with LCH from healthy controls, with an AUC of >86%, showing promising diagnostic potential (Fig. 6C).

Fig. 6. L-kynurenine and betaine aldehyde levels in patients with LCH and healthy individuals were determined by LC-MS/MS. (A) Betaine aldehyde levels in pediatric patients with LCH and controls. (B) L-kynurenine levels in pediatric patients with LCH and controls. (C) ROC curve for betaine aldehyde in pediatric patients with LCH. (D) Betaine aldehyde concentrations in TLS and IS subtypes. (E) L-kynurenine levels in TLS and IS groups. (F) L-kynurenine levels in various clinical subtypes of pediatric LCH. (G) L-kynurenine levels in pediatric patients with LCH with different prognoses. (H) Effect of L-kynurenine levels on event-free survival rates in pediatric patients with LCH. Patients were stratified into high- and low-level groups based on median L-kynurenine levels. **P<0.01, ***P<0.001.

Fig. 6

Open in a new tab

Abbreviations: LCH, Langerhans cell histiocytosis; TLS, tumor-like symptoms; IS, inflammatory symptoms; SS, single-system disease; MS-RO–, multisystem disease without risk organ involvement; MS-RO+, multisystem disease with risk organ involvement; GPG, good prognosis group; PPG, poor prognosis group.

Analysis across LCH subtypes revealed subtype-specific metabolic alterations. While betaine aldehyde levels did not significantly differ between the TIS and IS subtypes (Fig. 6D), L-kynurenine levels were significantly elevated in TIS compared with IS samples (P<0.001, Fig. 6E). L-kynurenine levels were similar between SS and MS-RO– patients but were significantly higher in MS-RO+ patients (P<0.001, Fig. 6F). Similarly, the levels were significantly elevated in PPG relative to GPG (P<0.001, Fig. 6G). Notably, L-kynurenine levels >3 µM (i.e., the median value) in patients with LCH were predictive of a reduced EFS rate (Fig. 6H) in this cohort, corroborating the results of untargeted metabolomics analysis.

DISCUSSION

Metabolomics is recognized as the closest molecular readout of phenotypes, bridging genetic perturbations and physiological manifestations [17, 18]. However, metabolic alterations in LCH remain unclear. We identified key metabolites in BMA for the diagnosis, prognosis, and therapeutic monitoring of pediatric LCH via metabolomic profiling, expecting to provide novel insights for clinical applications.

Besides global alterations, we identified a panel of nine strongly altered metabolites, including amino acids/peptides and fatty acids, relevant to the diagnosis of LCH in pediatric patients. Notably, oleamide, DTA, and myo-inositol, which we identified as potential diagnostic metabolites for LCH, accumulate abnormally in Alzheimer’s [19–21]. Progressive neurodegeneration is a severe complication of LCH, corroborating the reliability of our findings. Imbalances in the synthesis and transport of myo-inositol create dependency on this nutrient in other hematologic malignancies, such as acute myeloid leukemia (AML) [22]. Xanthine and hypoxanthine are important nucleic acid metabolic intermediates. Elevated hypoxanthine is a pan-cancer biomarker [23]. Nicotinamide (NAM) is a precursor to NAM-adenine dinucleotide, a key regulator of cellular energy metabolism, and accumulating evidence suggests that it plays a significant role in cancer prevention and therapy [24]. These metabolites are implicated in the occurrence and progression of diverse diseases. To the best of our knowledge, links between these specific metabolites and LCH have not yet been reported, warranting further investigation.

LCH is a complex disorder, particularly in subtypes with multisystem or internal organ/system involvement. Although gold-standard pathological criteria for the assessment of CD1A and/or CD207 are available, biopsies of the pituitary gland, vertebral bodies, and liver/spleen present more disadvantages than advantages and may hinder rapid classification. We identified various lipid and lipid-like metabolites that differed in level between the SS and MS subtypes. In particular, in MS LCH cases, several fatty acid metabolites were markedly decreased, whereas glycerophospholipids were significantly increased. The glycerophosphocholine metabolite PC(17:0/0:0) was gradually reduced in level across the three clinical subtypes (SS>MS-RO–>MS-RO+) and therefore holds potential as a molecular typing marker. Elevated levels of PC and choline-containing compounds have been associated with carcinogenesis [25]. However, as a major phospholipid component of biological membranes, PC may interact with C-reactive protein and thus contribute to host defense and inflammation modulation [26]. Given that LCH is an inflammatory disorder, PC may serve as a marker to distinguish the three subtypes of LCH.

Patients with LCH are traditionally clinically classified and treated based on organ involvement. Drawing on extensive clinical experience, our team has proposed an alternative approach to determine treatment based on the nature of clinical symptoms (TLS vs. IS), yielding highly favorable outcomes [14]. However, the mechanisms underlying symptoms remain unclear. We found that patients with TLS exhibited not only lipid and lipid-like metabolic abnormalities but also significant disturbances in benzenoids, organic acids and derivatives, and organ heterocyclic compounds, which were consistently highly elevated in patients with TLS. L-kynurenine and betaine aldehyde levels helped in effectively distinguishing patients with TLS from those with IS. Particularly, L-kynurenine, an immunosuppressive metabolite produced from L-tryptophan by indoleamine 2,3-dioxigenase 1 catalysis [27], served as a significant biomarker for poor prognosis in LCH. L-kynurenine contributes to cancer immune evasion and independently predicts poor overall survival in peripheral T-cell lymphoma and AML [28, 29]. Reports on betaine aldehyde in cancers are limited; however, it is a crucial intermediate in choline metabolism that intricately interacts with other metabolic pathways, including glycine, cysteine, methionine, and purine metabolism [30]. Assessing betaine aldehyde levels helps in distinguishing colorectal cancer tissues from adjacent mucosal tissues [31]. These findings indicate that both L-kynurenine and betaine aldehyde may play a key role in LCH classification based on symptoms such as inflammation and TLSs.

Despite incremental progress, evidence supporting the efficacy of specific therapies for severe LCH remains insufficient. A prospective blinded study revealed that plasma and urinary copy numbers of cell-free DNA harboring the BRAFV600E variant may serve as convenient and reliable indicators for monitoring therapeutic response in LCH related to BRAF variants [32]. We detected dynamic changes in four metabolites in BMA over the treatment course, irrespective of BRAF variants. Notably, inosine plays crucial roles as both an essential metabolite involved in purine biosynthesis and degradation and as a bioactive molecule regulating RNA editing, metabolic enzyme activity, and signaling pathways [33]. Inosine also acts as a critical factor that influences the response to immune checkpoint inhibition therapy in various cancers [34]. Moreover, dodecanoic acid levels are decreased in lung adenocarcinoma tissues [35], and urinary levels discriminate breast invasive ductal carcinoma from healthy controls [36]. In a metabolomics study of cerebrospinal fluid from pituitary stalk lesions, including brain-involved LCH, 29 of 67 metabolites were strongly associated with clinical status [37]. Metabolic panel monitoring during treatment may provide valuable information on the outcomes and efficacy of specific treatments. As Langerhans cells are epidermal dendritic cells derived from multipotent bone marrow stem cells, metabolites in BMAs may be an ideal alternative.

L-kynurenine may serve as a baseline LCH risk assessment tool; our data suggest that patients with L-kynurenine levels 3 μM have significantly poorer outcomes. In such high-risk cases, clinicians should consider more intensive first-line regimens (e.g., prolonged combination chemotherapy). Dodecanoic acid and inosine are promising for real-time therapeutic efficacy assessment, as their levels were highly elevated in patients with ND MS-RO+ LCH but normalized upon achieving CR. These metabolites can be combined with radiographic imaging and circulating BRAFV600E detection for multidimensional assessment.

We acknowledge that the proposed thresholds are preliminary and require validation in larger cohorts before clinical application. Nevertheless, our results provide a foundation for further studies, including the standardization of assay protocols across institutions, definition of age- and organ-specific reference ranges for these metabolites, and exploration of point-of-care adaptations for rapid monitoring. Other limitations of our study include the fact that the data were collected from a single center; future studies should include larger and multicenter cohorts. Further, gene–metabolite associations should be analyzed to uncover potential mechanisms of metabolic reprogramming. Finally, as LCH lesions can affect various organs, differences in metabolomic changes across tissues remain to be elucidated.

In conclusion, we systematically investigated BMA samples from pediatric patients with various LCH subtypes. Nine metabolites were identified as biomarkers with excellent diagnostic potential. PC(17:0/0:0), L-kynurenine, and betaine aldehyde have clinical significance for classification, and L-kynurenine potentially predicts prognosis in pediatric patients with LCH. A panel of metabolites, including dodecanoic acid and inosine, may serve as indicators for the assessment of treatment efficacy in MS-RO+ LCH, irrespective of BRAF variant status. Our results provide a comprehensive view of metabolite alterations in pediatric patients with LCH and may offer a foundation for improving diagnosis, prognosis, classification, and treatment. Further large-cohort, multicenter studies are warranted.

ACKNOWLEDGEMENTS

None.

SUPPLEMENTARY MATERIALS

Supplementary materials can be found via https://doi.org/10.3343/alm.2025.0536.

Footnotes

AUTHOR CONTRIBUTIONS

Liu XD and Mao XL conceptualized the study; Wang YL, Wang JY, and Lin HL contributed to methodology; Liu XD, Wang YL, Wang JY, Lin HL, Zheng QQ, and Cai MS contributed to investigation; Liu XD and Mao XL were responsible for visualization; Liu XD, Jiang H, and Mao XL acquired funding; Liu XD administrated the project; Cao XX, Jiang H, and Mao XL supervised the study; Liu XD wrote the original manuscript draft; Mao XL contributed to writing – review & editing. All authors have read and approved the final manuscript.

CONFLICTS OF INTEREST

None declared.

RESEARCH FUNDING

This study was in part supported by the National Key Research and Development Program of China (Grant No. 2022YFC2705003), the National Natural Science Foundation of China (Grant No. 82170176), and Guangzhou Medical University Discipline Construction Funds (Basic Medicine) (Grant No. JCXKJS2022A05).

REFERENCES

  • 1.Rodriguez-Galindo C, Allen CE. Langerhans cell histiocytosis. Blood. 2020;135:1319–31. doi: 10.1182/blood.2019000934. [DOI] [PubMed] [Google Scholar]
  • 2.Krooks J, Minkov M, Weatherall AG. Langerhans cell histiocytosis in children: history, classification, pathobiology, clinical manifestations, and prognosis. J Am Acad Dermatol. 2018;78:1035–44. doi: 10.1016/j.jaad.2017.05.059. [DOI] [PubMed] [Google Scholar]
  • 3.Chakraborty R, Burke TM, Hampton OA, Zinn DJ, Lim KPH, Abhyankar H, et al. Alternative genetic mechanisms of BRAF activation in Langerhans cell histiocytosis. Blood. 2016;128:2533–37. doi: 10.1182/blood-2016-08-733790. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Yang Y, Wang C, Wang D, Cui L, Li N, Lian H, et al. Clinical study of MAP2K1-mutated Langerhans cell histiocytosis in children. J Cancer Res Clin Oncol. 2022;148:2517–27. doi: 10.1007/s00432-021-03810-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Finley LWS. What is cancer metabolism? Cell. 2023;186:1670–88. doi: 10.1016/j.cell.2023.01.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Qiu S, Cai Y, Yao H, Lin C, Xie Y, Tang S, et al. Small molecule metabolites: discovery of biomarkers and therapeutic targets. Signal Transduct Target Ther. 2023;8:132. doi: 10.1038/s41392-023-01399-3.1a7077d173ec4d03acc39ed4585d5349 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Faubert B, Solmonson A, DeBerardinis RJ. Metabolic reprogramming and cancer progression. Science. 2020;368:eaaw5473. doi: 10.1126/science.aaw5473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Lin J, Xia L, Liang J, Han Y, Wang H, Oyang L, et al. The roles of glucose metabolic reprogramming in chemo- and radio-resistance. J Exp Clin Cancer Res. 2019;38:218. doi: 10.1186/s13046-019-1214-z.8b04269ded8742ac9f1b4c9ce32a60ca [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Zhang T, Zhang A, Qiu S, Yang S, Wang X. Current trends and innovations in bioanalytical techniques of metabolomics. Crit Rev Anal Chem. 2016;46:342–51. doi: 10.1080/10408347.2015.1079475. [DOI] [PubMed] [Google Scholar]
  • 10.Ronkina N, Gaestel M. MAPK-activated protein kinases: servant or partner? Annu Rev Biochem. 2022;91:505–40. doi: 10.1146/annurev-biochem-081720-114505. [DOI] [PubMed] [Google Scholar]
  • 11.Brummer C, Faerber S, Bruss C, Blank C, Lacroix R, Haferkamp S, et al. Metabolic targeting synergizes with MAPK inhibition and delays drug resistance in melanoma. Cancer Lett. 2019;442:453–63. doi: 10.1016/j.canlet.2018.11.018. [DOI] [PubMed] [Google Scholar]
  • 12.Hutton JE, Wang X, Zimmerman LJ, Slebos RJC, Trenary IA, Young JD, et al. Oncogenic KRAS and BRAF drive metabolic reprogramming in colorectal cancer. Mol Cell Proteomics. 2016;15:2924–38. doi: 10.1074/mcp.M116.058925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Waldschmidt JM, Kloeber JA, Anand P, Frede J, Kokkalis A, Dimitrova V, et al. Single-cell profiling reveals metabolic reprogramming as a resistance mechanism in BRAF-mutated multiple myeloma. Clin Cancer Res. 2021;27:6432–44. doi: 10.1158/1078-0432.CCR-21-2040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Lin HL, Zheng QQ, Huang RL, Hu R, Liu XD, Wang JY. Exploration of treatment in childhood Langerhans cell histiocytosis based on inflammatory and malignant symptoms: a pilot study. Orphanet J Rare Dis. 2024;19:174. doi: 10.1186/s13023-024-03151-8.5b38e24ebc354fd8823da4f5a85e8cf2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Li Z, Zhang T, Xu L, Wei Y, Cui H, Tang Y, et al. Plasma metabolic alterations and potential biomarkers in individuals at clinical high risk for psychosis. Schizophr Res. 2022;239:19–28. doi: 10.1016/j.schres.2021.11.011. [DOI] [PubMed] [Google Scholar]
  • 16.Blaženović I, Kind T, Ji J, Fiehn O. Software tools and approaches for compound identification of LC-MS/MS data in metabolomics. Metabolites. 2018;8:31. doi: 10.3390/metabo8020031.ab74ece87082453a86fe96b054ec1287 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zhang A, Sun H, Xu H, Qiu S, Wang X. Cell metabolomics. OMICS. 2013;17:495–501. doi: 10.1089/omi.2012.0090. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Johnson CH, Ivanisevic J, Siuzdak G. Metabolomics: beyond biomarkers and towards mechanisms. Nat Rev Mol Cell Biol. 2016;17:451–59. doi: 10.1038/nrm.2016.25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Gómez-Pascual A, Naccache T, Xu J, Hooshmand K, Wretlind A, Gabrielli M, et al. Paired plasma lipidomics and proteomics analysis in the conversion from mild cognitive impairment to Alzheimer's disease. Comput Biol Med. 2024;176:108588. doi: 10.1016/j.compbiomed.2024.108588. [DOI] [PubMed] [Google Scholar]
  • 20.Lohner S, Vágási J, Marosvölgyi T, Tényi T, Decsi T. Inverse association between 18-carbon trans fatty acids and intelligence quotients in smoking schizophrenia patients. Psychiatry Res. 2014;215:9–13. doi: 10.1016/j.psychres.2013.10.011. [DOI] [PubMed] [Google Scholar]
  • 21.Song T, Song X, Zhu C, Patrick R, Skurla M, Santangelo I, et al. Mitochondrial dysfunction, oxidative stress, neuroinflammation, and metabolic alterations in the progression of Alzheimer's disease: a meta-analysis of in vivo magnetic resonance spectroscopy studies. Ageing Res Rev. 2021;72:101503. doi: 10.1016/j.arr.2021.101503. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Wei Y, Huang YH, Skopelitis DS, Iyer SV, Costa ASH, Yang Z, et al. SLC5A3-dependent myo-inositol auxotrophy in acute myeloid leukemia. Cancer Discov. 2022;12:450–67. doi: 10.1158/2159-8290.CD-20-1849. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Yu M, Liu C, Cao M, Yang D, Wang T, Xu J, et al. Purine metabolism: a pan-cancer metabolic dysregulation across circulation and tissues. Mol Cancer. 2025;24:255. doi: 10.1186/s12943-025-02482-9.67f4027f2c8449749a71da1d698fbc51 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Nikas IP, Paschou SA, Ryu HS. The role of nicotinamide in cancer chemoprevention and therapy. Biomolecules. 2020;10:477. doi: 10.3390/biom10030477.4453a9cf37a349b4b254f760ed8b3c57 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Glunde K, Jacobs MA, Bhujwalla ZM. Choline metabolism in cancer: implications for diagnosis and therapy. Expert Rev Mol Diagn. 2006;6:821–9. doi: 10.1586/14737159.6.6.821. [DOI] [PubMed] [Google Scholar]
  • 26.Richter K, Mathes V, Fronius M, Althaus M, Hecker A, Krasteva-Christ G, et al. Phosphocholine - an agonist of metabotropic but not of ionotropic functions of α9-containing nicotinic acetylcholine receptors. Sci Rep. 2016;6:28660. doi: 10.1038/srep28660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Adams S, Braidy N, Bessede A, Brew BJ, Grant R, Teo C, et al. The kynurenine pathway in brain tumor pathogenesis. Cancer Res. 2012;72:5649–57. doi: 10.1158/0008-5472.CAN-12-0549. [DOI] [PubMed] [Google Scholar]
  • 28.Shibata Y, Hara T, Matsumoto T, Nakamura N, Nakamura H, Ninomiya S, et al. Serum concentrations of L-kynurenine predict clinical outcomes of patients with peripheral T-cell lymphoma, not otherwise specified. Hematol Oncol. 2017;35:637–44. doi: 10.1002/hon.2318. [DOI] [PubMed] [Google Scholar]
  • 29.Mabuchi R, Hara T, Matsumoto T, Shibata Y, Nakamura N, Nakamura H, et al. High serum concentration of L-kynurenine predicts unfavorable outcomes in patients with acute myeloid leukemia. Leuk Lymphoma. 2016;57:92–8. doi: 10.3109/10428194.2015.1041388. [DOI] [PubMed] [Google Scholar]
  • 30.Guo K, Cao Y, Li Z, Zhou X, Ding R, Chen K, et al. Glycine metabolomic changes induced by anticancer agents in A549 cells. Amino Acids. 2020;52:793–809. doi: 10.1007/s00726-020-02853-0. [DOI] [PubMed] [Google Scholar]
  • 31.Brown DG, Rao S, Weir TL, O'Malia J, Bazan M, Brown RJ, et al. Metabolomics and metabolic pathway networks from human colorectal cancers, adjacent mucosa, and stool. Cancer Metab. 2016;4:11. doi: 10.1186/s40170-016-0151-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Hyman DM, Diamond EL, Vibat CRT, Hassaine L, Poole JC, Patel M, et al. Prospective blinded study of BRAFV600E mutation detection in cell-free DNA of patients with systemic histiocytic disorders. Cancer Discov. 2015;5:64–71. doi: 10.1158/2159-8290.CD-14-0742. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Kim IS, Jo EK. Inosine: a bioactive metabolite with multimodal actions in human diseases. Front Pharmacol. 2022;13:1043970. doi: 10.3389/fphar.2022.1043970.57974632ab474c2cb6690f1180483889 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Samami E, Aleebrahim-Dehkordi E, Mohebalizadeh M, Yaribash S, Saghazadeh A, Rezaei N. Inosine, gut microbiota, and cancer immunometabolism. Am J Physiol Endocrinol Metab. 2023;324:E1–E8. doi: 10.1152/ajpendo.00207.2022. [DOI] [PubMed] [Google Scholar]
  • 35.Zhang X, Tong X, Chen Y, Chen J, Li Y, Ding C, et al. A metabolomics study on carcinogenesis of ground-glass nodules. Cytojournal. 2024;21:12. doi: 10.25259/Cytojournal_68_2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Taunk K, Taware R, More TH, Porto-Figueira P, Pereira JAM, Mohapatra R, et al. A non-invasive approach to explore the discriminatory potential of the urinary volatilome of invasive ductal carcinoma of the breast. RSC Adv. 2018;8:25040–50. doi: 10.1039/C8RA02083C. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Wang L, Chen X, Liu X, Miao H, Gong F, Yang H, et al. Cerebrospinal fluid metabolomic pattern of different pituitary stalk lesions. J Clin Endocrinol Metab. 2024;109:802–14. doi: 10.1210/clinem/dgad559. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

alm-46-5-539-supple.pdf (1.3MB, pdf)

Articles from Annals of Laboratory Medicine are provided here courtesy of Korean Society for Laboratory Medicine

RESOURCES