Gut microbiome modulates the outcome in primary central nervous system lymphoma patients undergoing chemotherapy: An ancillary study from the BLOCAGE trial

Isaias Hernández-Verdin; Eva Kirasic; Karima Mokhtari; Noemie Barillot; Lucas Rincón de la Rosa; Elise Sourdeau; Yahse Abada; Magali Le Tarff-Tavernier; Lucia Nichelli; Laura Rozenblum; Aurélie Kas; Bertrand Mathon; Sylvain Choquet; Caroline Houillier; Khê Hoang-Xuan; Agusti Alentorn

doi:10.1093/neuonc/noaf059

. 2025 Mar 6;27(8):2090–2104. doi: 10.1093/neuonc/noaf059

Gut microbiome modulates the outcome in primary central nervous system lymphoma patients undergoing chemotherapy: An ancillary study from the BLOCAGE trial

Isaias Hernández-Verdin ¹, Eva Kirasic ², Karima Mokhtari ^3,⁴, Noemie Barillot ⁵, Lucas Rincón de la Rosa ⁶, Elise Sourdeau ⁷, Yahse Abada ⁸, Magali Le Tarff-Tavernier ⁹, Lucia Nichelli ¹⁰, Laura Rozenblum ¹¹, Aurélie Kas ¹², Bertrand Mathon ¹³, Sylvain Choquet ¹⁴, Caroline Houillier ¹⁵, Khê Hoang-Xuan ^16,¹⁷, Agusti Alentorn ^18,^19,^✉

¹ Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

² Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

³ Department of Neuropathology, Groupe Hospitalier Pitié Salpêtrière, APHP, Paris, France

⁴ Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

⁵ Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

⁶ Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

⁷ Department of Hematology, APHP, Hôpital Pitié-Salpêtrière and Sorbonne University, Paris, France

⁸ Department of Neuroradiology, Groupe Hospitalier Pitié Salpêtrière, APHP, Paris, France

⁹ Department of Hematology, APHP, Hôpital Pitié-Salpêtrière and Sorbonne University, Paris, France

¹⁰ Department of Neuroradiology, Groupe Hospitalier Pitié Salpêtrière, APHP, Paris, France

¹¹ Sorbonne Université, AP-HP, Hôpitaux Universitaires Pitié-Salpêtrière Charles Foix, Service de Médecine Nucléaire and LIB, INSERM U1146, Paris, France

¹² Sorbonne Université, AP-HP, Hôpitaux Universitaires Pitié-Salpêtrière Charles Foix, Service de Médecine Nucléaire and LIB, INSERM U1146, Paris, France

¹³ Department of Neurosurgery, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France

¹⁴ Department of Hematology, APHP, Hôpital Pitié-Salpêtrière and Sorbonne University, Paris, France

¹⁵ Department of Neurology-2, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France

¹⁶ Department of Neurology-2, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France

¹⁷ Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

¹⁸ Department of Neurology-2, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France

¹⁹ Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France

^✉

Corresponding Author: Agustí Alentorn, MD, PhD, Department of Neurology-2, Mazarin, Groupe Hospitalier Pitié Salpêtrière, 75013 Paris, France (agusti.alentorn@icm-institute.org).

PMCID: PMC12448904 PMID: 40048707

Abstract

Background

Primary central nervous system lymphoma (PCNSL) treatment relies on a high-dose methotrexate-based chemotherapy (HD-MTX-based CT) regimen; however, whether there is a specific microbiota composition association with treatment response and clinical outcomes remains incompletely understood.

Methods

We conducted a prospective study of PCNSL patients, included in the clinical trial NCT02313389 and the ancillary study NCT04253496 from 2020 to 2023, where patients were treated with first-line HD-MTX-based polychemotherapy without a consolidation treatment. Stool (n = 52), cerebrospinal fluid (CSF, n = 52), and plasma samples (n = 35) were collected before and/or after therapy initiation to perform metagenomic, flow cytometry, and metabolomic analyses. Plasma metabolomic data of 90 patients also included in the BLOCAGE clinical trial was subsequently used as a validation cohort.

Results

Unsupervised clustering of microbial data identified two distinct gut microbial communities, differing in Parabacteroides distasonis abundance, which correlated with progression-free survival and overall survival in both uni- and multivariate analyses. Higher P. distasonis levels were linked to increased plasma betaine–valine metabolites and enhanced CD8 T cell infiltration in the CSF, suggesting a connection between gut microbiota and immune regulation. Stratifying the validation cohort by betaine–valine content confirmed these clinical associations.

Conclusions

Our findings suggest that gut microbiome communities modulate clinical outcomes in PCNSL patients undergoing standard treatment. Moreover, after future validation in external cohorts, the quantification of Parabacteroides distasonis could potentially provide a basis for patient stratification and guide personalized therapeutic strategies in the near future.

Keywords: clinical outcomes, gut microbiome, metabolomics, PCNSL

Key Points.

Two microbial communities are associated with outcomes in HD-MTX-treated PCNSL patients.
Parabacteroides distasonis content associates CD8 T cell infiltration to the cerebrospinal fluid.
Plasma betaine–valine content associatesd with Parabacteroides distasonis presence.

Importance of the Study.

This study reveals a novel link between Parabacteroides distasonis and treatment response to high-dose methotrexate-based chemotherapy (HD-MTX-based CT) in primary central nervous system lymphoma (PCNSL) patients. Higher P. distasonis content was associated with increased CD8 T cell infiltration into the cerebrospinal fluid (CSF), indicating immune modulation via the gut-brain axis. Furthermore, patients with higher plasma levels of the betaine–valine metabolite, correlated with P. distasonis abundance, exhibited greater infiltration of macrophages and exhausted CD8 T cells into the tumor. Prior research has mainly focused on other cancer types, making this the first comprehensive investigation linking the gut microbiome to HD-MTX-based CT response in PCNSL. These findings highlight the potential impact of gut microbial communities on immune responses in PCNSL and underscore their translational potential for microbiome-based patient stratification. Future research should focus on validating these findings and exploring microbiome-targeted interventions, such as prebiotics or probiotics, to enhance treatment efficacy.

Primary central nervous system lymphoma (PCNSL) is a rare subtype of extranodal non-Hodgkin’s lymphoma (NHL) within diffuse large B cell lymphoma (DLBCL) arising exclusively within the brain, spinal cord, leptomeninges, or eye.¹ Despite PCNSL being a rare entity compared to other systemic lymphomas, their incidence has progressively increased during the last 2 decades, especially in patients aged 70–79 years.^2–6 PCNSL are associated with dismal and heterogeneous prognosis, with a median overall survival (OS) and 5-year survival of 26 months and 22%, respectively.^7,8 The standard treatment relies on high-dose methotrexate-based chemotherapy (HD-MTX-based CT) regimen with or without consolidation; however, despite PCNSL being chemosensitive, 33% of the patients are refractory to first-line treatment, and up to 60% of the patients will eventually relapse after 2–5 years of the initial diagnosis.^7,9 To investigate other treatment strategies, the phase III randomized controlled trial (BLOCAGE01 NCT02313389), evaluated the benefits (prolonged complete remission) of maintenance polychemotherapy (MTX, rituximab, and temozolomide) versus observation in elderly patients with PCNSL after standard induction HD-MTX-based CT (MTX, rituximab, procarbazine, vincristine, and cytarabine).

Molecularly, PCNSL is characterized by constitutive Nuclear Factor-κB (NF-κB) activity, which leads to cell proliferation and prevention of cellular apoptosis, driven by alterations in genes of the BCR pathway (CD79B in 43%), of the toll-like receptor (TLR) pathway (MYD88 L265P in 64%) among others.^10–13 Moreover, the tumor microenvironment (TME) of these tumors is highly heterogeneous and presents a strong relationship with the clinical outcome.^14–16

Heterogeneity beyond tumor genomics and the TME include host factors such as the gut microbiome which has gained attention since it affects cancer development and therapeutic responses.^17–21 The gut microbiome can modulate the immune system indirectly through the production of specific metabolites,²² ultimately affecting treatment response in various cancer types, including lymphomas.^{17–21,23–25} In PCNSL, today, there is only one study attempting to characterize the gut microbiome composition difference between PCNSL patients and healthy donors²⁶; however, there is no study performing correlations with clinical variables. To investigate the role of intestinal microbiota in the treatment response and clinical outcomes in HD-MTX-based CT-treated immunocompetent Epstein-Barr negative (EBV-) PCNSL patients, included in the BLOCAGE clinical trial (NCT02313389) and the ancillary study NCT04253496, we prospectively collected gut microbiome, cerebrospinal fluid (CSF), and plasma-derived metabolomic data from PCNSL patients before and/or after treatment.

Methods

Patients

A total of 52 fecal stool samples from 43 treatment-naive immunocompetent Epstein-Barr negative PCNSL patients, collected at the Pitié Salpêtirère Hospital (France) and included in the BLOCAGE01 study, were eligible to this prospective ancillary study. The inclusion criteria included (1) newly diagnosed PCNSL, confirmed either by brain histology or cytology in the cerebrospinal fluid (CSF) or vitrectomy; (2) age of 60 years or older; (3) Karnofsky Performance Status (KPS) greater than 40; and (4) no evidence of systemic non-Hodgkin lymphoma as demonstrated by a contrast-enhanced CT and/or whole-body [18F]FDG PET/CT scan.⁷ Exclusion criteria included (1) any other active malignancy (with the exception of basal cell carcinoma of the skin and cervical carcinoma in situ), (2) pre-existing immunodeficiency (from HIV or organ transplant treatment), (3) low-grade histology, (4) isolated primary vitreoretinal lymphoma, (5) central nervous system (CNS) relapse of a systemic lymphoma, and (6) prior treatment for PCNSL. Patients received a HD-MTX-based CT regimen which consisted of four 28-day cycles of MTX (3.5 g/m², D1 and D15), rituximab (375 mg/m², D1), procarbazine (100 mg/ m², D1 to D7), and vincristine (1.4 mg/m², D1 and D15; R-MPV) followed by a cycle of cytarabine (3 g/m², D1, D2). Only patients who obtained a complete response (CR) or unconfirmed complete response (CRu) were subsequently randomized into 2 groups in the BLOCAGE01 study: maintenance arm with rituximab, MTX, and temozolomide or observation arm (wait and see [WS]). The remaining patients were not randomized. Therapeutic response was assessed at end of therapy by board-certified neurooncologists and neuroradiologists (LN) using International PCNSL Collaborative Group criteria (IPCG), based on MR imaging, corticosteroid use, and CSF cytology and slit lamp examination in case of CSF or ocular involvement at baseline. CR requires the complete disappearance of all signs of the disease on gadolinium-enhanced MRI as well as on CSF cytology and ocular examination. CRu includes patients who meet CR criteria but continue to require corticosteroids or show minor abnormalities. All patients received corticosteroids after diagnosis confirmation. Patients were not subjected to surgical resection, but stereotactic biopsy was performed in all the analyzed patients.

Progression-free survival (PFS) was defined as the time between the diagnosis and the progression of the disease or the death of the patient. OS was calculated as the time between the diagnosis and the death of the patient whatever the cause.

We performed metagenomic shotgun sequencing on 52 fecal stool samples (n = 52), flow cytometry (FC) from 52 CSF samples, and metabolomics from 90 plasma samples, all included in the BLOCAGE clinical trial. 35 patients had all data sources while 55 only had metabolomic data. From the 52 patients’ fecal stool samples, 35 had only one sampling time, 7 had 2, and 1 had 3. Six patients received antibiotics treatment before the first sample collection (one fluconazole, four sulfamethoxazole + trimethoprim, and one sulfamethoxazole + trimethoprim + fungizone), but not at the second nor the third. RNA sequencing (RNA-seq, n = 28) and whole exome sequencing (WES, n = 27) were recovered from a previous already published work.¹³ The overall study design is depicted in Figure 1.

Study design and data flowchart. Panel A depicts the study workflow, while Panel B shows data composition per cohort. Microbiome sequencing was conducted on 52 fecal samples from 43 patients, with 8 having a second sampling and 1 having a third. Validation was performed using metabolomics on 90 plasma samples, 35 of which had corresponding microbiome data. — Study design and flowchart. (A) The study design and flowchart (B) indicate the data composition for each cohort. Microbiome sequencing was performed on 52 fecal samples, 43 unique patients with at least 1 sampling point, 8 with a second sampling, and 1 with a third sampling. Validation was performed indirectly through metabolomics on 90 plasma samples from which 35 had the microbiome data.

Fecal Microbiome Collection and Processing

The fecal sample collection occurred from 2018 to 2021. Samples were stored at −40 °C within 6 hours of collection. All patients consented to biospecimen protocols included in the BLOCAGE clinical trial (NCT02313389) and the ancillary study NCT04253496. From the first time of sample collection (n = 43), 72.1% (31/43) samples were collected before the very first chemotherapy treatment (median = 3 days before chemotherapy; CI_95% = 1–4 days before), 25.6% (11/43) after the very first chemotherapy treatment (median = 7 days after chemotherapy; CI_95% = 1–909 days after), and 2.3% (1/43) samples were unknown (Figure 2). The second and third sample collections were all taken after treatment (median = 302 days after the first chemotherapy; CI_95% = 114–454 days after). Microbiome DNA was extracted using the Maxwell RSC Fecal Microbiome DNA Kit (Maxwell cat # AS1700).

Bar plot displaying the phylogenetic composition of fecal samples (n = 52) from primary central nervous system lymphoma patients. Samples are ordered by decreasing inverse Simpson alpha diversity values and grouped by patients with 1 sampling point (n = 35), 2 sampling points (n = 7), and 3 sampling points (n = 1). — Distribution of microbiome composition in primary central nervous system lymphoma patients. Barplot showing the phylogenetic composition of fecal samples (n = 52). Samples are ordered by decreasing inverse Simpson alpha diversity values and by a group of patients having only 1 sampling point (n = 35, left), 2 sampling points (n = 7), or 3 sampling points (n = 1).

Metagenomic Shotgun Sequencing and Analysis

Extracted DNA was then prepared for sequencing using the Illumina DNAprep (M) library preparation kit and sequenced on an illumina Nova-seq 6000 sequencing system targeting approximately 162.5 million reads (CI_95% = 157.9–168.6) per sample with 2 × 151 bp paired-end sequencing. Removal of human (GRCh38) contamination was done using Bowtie2 (v2.5.1) with the paired-end reads.²⁷ After decontamination, the paired-end reads were used for taxonomic profiling by Kraken 2 (v2.1.2) using a 62 Gb database (https://genome-idx.s3.amazonaws.com/kraken/k2_pluspf_20221209.tar.gz) with default parameters (except for “--minimum-hit-groups” which was set to 3).^28,29 Bracken (v2.6.0) was used for abundance estimation using the parameter”−r 50 −t 10.”³⁰ Kraken reports were combined into one “biom” file using kraken-biom (v1.0.1). In parallel, decontaminated paired-end reads were concatenated to a single FASTQ file as the input for functional profiling with the HUMAnN v3.8 (http://huttenhower.sph.harvard. edu/humann) pipeline.³¹ After aligning to the updated MetaPhlAn (v3.0) and UniRef90 database (default settings), and MetaCyc version 24.0 was used to obtain stratified and unstratified pathway abundances.³² HUMAnN’s default RPK values on abundances were renormalized by library depth to copies per million or to relative abundances using the “humann_renorm_table” function.

Phyloseq (v1.40.0) was used to import into R (v4.3.1) the “biom” file containing the count’s file and perform downstream analyses.³³ Low abundance taxa were filtered out (>10 raw reads). Inverse Simpson alpha, Chao1, and Shannon diversity were calculated on a previously rarefied counts table using “rarefy_even_depth” from phyloseq. ANOSIM and PERMANOVA (adonis function from vegan v2.6-4) were used to detect changes in community composition, while Beta dispersion tests (betadisper from vegan package) were used to check whether differences are due to group dispersion using counts from rarefied data and Bray-Curtis distances.³⁴ Principal coordinate analysis (PCoA) was performed using Canberra distance on centered log2-ratio (clr)²⁰ normalized counts where operational taxonomic units (OTUs) were filtered out if they had a relative abundance less than <0.01%. DESeq2 (v1.40.2) was used to determine differentially abundant taxa on raw count data.³⁵

Determination of common microbial communities across patients was performed on samples at first recollection (time point = 1) where three samples were dropped, one because it died to early (before HDMTX treatment initiation) and other 2 because of antifungal treatment (which was proven to affect the microbial composition in our analyses). Relative counts were first filtered for OTUS having a relative abundance less than <0.01%, then the resulting matrix was used as input for consensus unsupervised clustering resulting from 10 different clustering algorithms (iClusterBayes, moCluster, CIMLR, IntNMF, ConsensusClustering, COCA, NEMO, PINSPlus, SNF, and LRA).^36–46 The most appropriate cluster number was obtained by clustering prediction index (CPI) and Gaps-statistics analyses. Silhouette score was calculated to measure sample similarity across the detected clusters.⁴⁷ Most of the above analyses are integrated in the R package “MOVICS.”⁴⁶ The Linear Discriminant Analysis (LAD) effect size (LEfSe, v.1.1) and MaAsLin3 (Microbiome Multivariable Associations with Linear Models) algorithms^48,49 were employed to identify differentially abundant OTUS (relative abundance table) or functional pathways (stratified and unstratified) between groups (eg, between clusters/ using as subclass the “chemotherapy timing”). The LEfSe method first detected features with significant differential abundance using the non-parametric factorial Kruskal–Wallis rank-sum test with predefined alpha of 0.05. Significant features were then used to build a LDA model for estimating the effect size of each differentially abundant feature.

Metabolomics

All samples were taken before therapy initiation. Serum sample preparation and widely targeted detection by LC-MS. Fifty (50) µl of collected sera were mixed with 500 µl of ice-cold extraction mixture (methanol/water, 9/1, −20 °C, with labeled internal standard). To facilitate endogenous metabolites extraction, samples were then completely homogenized (vortexed 5 min at 2500 rpm) and then centrifuged (10 min at 15 000 g, 4 °C). Supernatants were collected and several fractions were split to be analyzed by different Liquid chromatography coupled with mass spectrometers (LC/MS).^50,51 Polyamines biliary acids and small/polar organic acids analyses were performed by LC-MS/MS with a 1290 UHPLC (Ultra-High Performance Liquid Chromatography; Agilent Technologies) coupled to a QQQ 6470 (Agilent Technologies).

Regarding polyamines by MRM analysis in positive polarity, the gas temperature was set to 350 °C with a gas flow of 12 l/min. The capillary voltage was set to 2.5 kV. Ten (10) μl of the sample were injected on a Column Kinetex C18 (150 × 2.1 mm particle size 2.6 µm) from Phenomenex, protected by a guard column C18 (5 × 2.1 mm) and heated at 40 °C by a Pelletier oven. The gradient mobile phase consisted of water with 0.1% of Heptafluorobutyric acid (HFBA, Sigma-Aldrich) (A) and acetonitrile with 0.1% of HFBA (B) freshly made. The flow rate was set to 0.4 ml/min, and gradient was follows: initial condition was 95% phase A and 5% phase B. Molecules were then eluted using a gradient from 5% to 30% phase B over 7 minutes. The column was washed using 90% mobile phase B for 2.25 minutes and equilibrated using 5% mobile phase B for 4 minutes. The autosampler was kept at 4 °C⁵⁰.

Regarding biliary acids by MRM analysis in negative polarity, the gas temperature was set to 310 °C with a gas flow of 9 L/min. The capillary voltage was set to 4.5 kV. Ten (10) μl of sample were injected on a Column Poroshell 120 EC-C8 1200bars (P/N 981758-902, 100× 2.1 mm particle size 1.9 µm) from Agilent technologies, protected by a guard column XDB-C18 (5 × 2.1 mm particle size 1.8 μm) and heated at 40 °C by a Pelletier oven. Gradient mobile phase consisted of water with 0.2% of formic acid (A) and acetonitrile/isopropanol (1/1; v/v) (B) freshly made. Flow rate was set to 0.5 mL/min, and gradient as follows: initial condition was 70% phase A and 30% phase B, changing to 38% phase B over 2 minutes. Phases proportion was still over 2 minutes, then molecules were eluted using a gradient from 38% to 60% phase B over half a minute. Column was washed using 98% mobile phase B for 2 minutes and equilibrated using 30% mobile phase B for 1.5 minutes. Autosampler was kept at 4°C⁵¹.

Regarding the small organic acid ketone bodies and polar metabolites (so-called SF method) by MRM analysis in both positive and negative polarity, alternatively, gas temperature was set to 300 °C with a gas flow of 12 L/min. Capillary voltage was set to 4 kV in positive and 5 kV in negative polarity mode. Ten (10) μl of sample were injected on a column Zorbax Eclipse XDB- (P/N 981758-902, 10 × 2.1 mm particle size 1.8 µm 1200 bars) from Agilent technologies, protected by a guard column XDB-C18 (5 × 2.1 mm particle size 1.8 μm) and heated at 50 °C by a Pelletier oven. Gradient mobile phase consisted of water with 0.1% of formic acid (A) and 0.1% of formic acid in acetonitrile (B) freshly made. Flow rate was set to 0.7 mL/min, and gradient as follows: initial condition was 80% phase A and 20% phase B for 3 minutes, changing to 45% phase B over 4 minutes before rinsing phase and equilibrium before next injection. Autosampler was kept at 4 °C.

Widely targeted by GC-MS/MS and pseudo-targeted analysis by UHPLC-HRAM (Ultra-High Performance Liquid Chromatography—High-Resolution Accurate Mass) was performed on a U3000 (Dionex)/Orbitrap q-Exactive (Thermo) coupling, previously described.^52,53

All targeted treated data were merged and cleaned with a dedicated R (version 4.0) package (@Github/Kroemerlab/GRMeta).

Flow Cytometry on CSF Samples

The CSF was obtained by lumbar puncture into four collection tubes (0.5–1 mL per tube) containing the TransFix® cell-stabilizing agent for FC. The mean time between sample collection and processing was around 4 hours but did not exceed 24 hours. A 6-color FC combination (CD45, CD3, CD4, CD8, CD5, CD19) was used to label the cell populations, which were then analyzed on a FACSCanto II cytometer (BD Biosciences). A sample was classified as FC⁺ when a cluster of at least 20 events with neoplastic features was detected. Clusters of 10 to 19 events with neoplastic features were classified as suspicious. Clusters with less than 10 events were classified as negative (FCM⁻).⁵⁴

Statistical Analyses

All statistical analyses were performed using the R statistical programming environment (v4.3.1). Differences in proportions and binary/categorical variables were calculated from two-sample Z-tests or Fisher’s exact test. Kruskal–Wallis test was used to test for a difference in distribution between three or more independent groups, and Mann–Whitney U test was used for differences in distributions between 2 population groups unless otherwise noted. P-values were corrected for multiple comparisons using the Benjamini-Hochberg method when applicable. OS and PFS analysis were assessed using log-rank Kaplan–Meier curves and multivariate Cox proportional hazards regression modeling.

Ethics Statement

All patients have provided written informed consent for their participation, and this study has been approved by a local institutional review board, included in the trials NCT02313389 and NCT04253496.

Results

Distribution of Gut Microbiota in PCNSL Patients

Organizing samples by decreasing inverse Simpson alpha diversity values, a metric that considers the number of unique organisms in a sample and the evenness with which they are distributed, showed that the baseline fecal samples had heterogeneous bacterial phylogenetic compositions, including high abundances of Bacteroides, Phocaeicola, Faecalibacterium, Ruminococcus along with lower but still heterogenous abundances of Escherichia, Verrucomicrobia, and the bacteriophage Uroviricota (Figure 2). Moreover, since we observed that the 2 patients (2105d95f and eb0ee378) taking antifungals had altered microbiota compositions, we wondered if antibacterials and/or antifungals intake before sampling could affect microbiota composition.

Antifungals Intake Before HD-MTX-based CT Treatment Affects Gut Microbiome in PCNSL Patients

As the primary objective of this study was to find associations between the baseline gut microbiome composition and the outcomes of HD-MTX-based CT treatment in PCNSL patients, we first needed to evaluate whether antibiotic intake affects the microbiome to avoid confounding variables. We found a trend of lower alpha diversity in patients taking antibiotics (n = 6) before fecal sampling versus those not taking antibiotics when using distinct ecological community metrics, including the inverse Simpson, Chao, and Shanon (Supplementary Figure 1A). Visualization of beta diversity on Canberra distances by PCoA revealed significant differences between patients taking antibiotics (P = .02, Supplementary Figure 1B). Moreover, PCoA also revealed that patients taking antifungals clustered further away, while those taking only antibacterials grouped with patients not undergoing antibiotics treatment (Supplementary Figure 1B). Given these results we decided to remove the patients under antifungal treatment before stool sampling from the subsequent analyses.

HD-MTX-based CT Treatment Depletes Akkermansia Muciniphila and Bifidobacterium Dentium Gut Microbiome Content in PCNSL Patients

Another potential confounding variable in our study was when the stool sample was taken, hence we compared paired fecal microbiome data of 6 PCNSL patients at baseline versus after the very first chemotherapy (patients c8ad5a74 and 43235177 dropped out since both time points were after their first chemotherapy, Figure 2). Inverse Simpson alpha diversity and PCoA analysis showed no significant difference in the microbiome diversity (Figure 3A and 3B). However, differential abundance analysis at species level revealed a depletion of Akkermansia muciniphila (Log₂FC = −6.018; P adjusted = .011), and Bifidobacterium dentium (Log₂FC = −3.500; P adjusted = .004; Figure 3C and Supplementary Figure 1C). Moreover, analysis using reconstructed metabolic pathways using HUMAnN and MetaCyc hierarchy of pathway classifications,^31,32 revealed depletion of pathways such as glycolysis, Coenzyme A, and ribonucleotide biosynthesis; whereas enrichment of fatty acid and beta-oxidation, isoleucine biosynthesis, mannitol cycle, among others (Figure 3D). Overall, this suggests that HD-MTX-based CT depletes specific microorganisms and affects metabolic activity; however, it does not broadly impact the microbiome composition in our cohort.

Impact of high-dose methotrexate-based chemotherapy on microbiome composition in primary central nervous system lymphoma patients. (A) Boxplot showing fecal diversity before and after the first chemotherapy. (B) PCoA plot using Canberra distance to compare samples pre- and post-chemotherapy. (C) Boxplot of differentially abundant OTUs between pre- and post-chemotherapy samples. (D) Top 25 differentially abundant functional pathways, with relative abundance and log2 fold changes annotated with adjusted P values. — High-dose methotrexate-based chemotherapy effect on microbiome composition in primary central nervous system lymphoma patients. (A) Boxplot Fecal diversity between samples of patients before or after the very first chemotherapy, calculated with inverse Simpson diversity index. (B) Principal coordinate analysis of fecal samples by chemotherapy (after vs. before) using Canberra distance. P value and R of the distance between groups were calculated by ADONIS (n = 12; 6 vs. 6), while the beta dispersion within groups was calculated with a permutation test. (C) Boxplot shows the differentially abundant OTUS comparing microbiome samples of patients after the first chemotherapy versus before chemotherapy. (D) Top 25 differentially abundant functional pathways between the microbiome of patients before versus after the very first chemotherapy. Left side shows the relative abundance per group. Right side the log2 fold changes with the adjusted P values annotated next to them. For (A) and (C), P value calculated by Wilcoxon matched-pairs signed rank test for paired data (per sample) The middle line is the median, the box limits represent the upper and lower quartiles, the whiskers note the 1.5× the iQr and the dots represent the individual data points.

Gut Microbiome Composition Impacts Outcome in PCNSL Patients Undergoing HD-MTX-based CT Treatment

Based on these data, we decided to use microbiome data from patients at the first recollection point except from 3 samples that were dropped, one because it died too early (before HD-MTX-based CT treatment initiation) and the other 2 because of antifungal treatment. The resulting 40 samples were used to determine common microbial communities across PCNSL patients, specifically, we used relative counts as input and performed consensus unsupervised clustering resulting from 10 different clustering algorithms. Using the optimal cluster number resulting from Consensus Partitioning Index (CPI), Gaps-statistics, and weighted silhouette width, we identified 2 microbial communities subtypes (CS1 and CS2, Figure 4A and Supplementary Figure 2A) that display different clinical outcomes in OS (Log-rank P = .004, 37.2 months vs. not-reached [NR], 95% confidence interval [CI] = 11.4–NR vs. NR) and PFS (Log-rank P = .001, 7.6 months vs. NR, 95% C.I. = 5.28–26.33 vs. NR, Figure 4B). After using Cox proportional hazard ratio multivariate models adjusting by age as a continuous variable, KPS as binary, sex, and the arm of randomization, the observations remained significant for the PFS (Figure 4C and Supplementary Figure 2B). There were no significant differences in the distribution of samples taken after or before the first chemotherapy, Firmicutes to Bacteroidetes ratio, nor higher height or body mass index (BMI) between groups (Supplementary Table 1). Moreover, in spite of not observing differences in the alpha diversity between communities, there was a significant difference in the microbiome composition as illustrated in the PCoA plot (Figure 4D and 4E).

Gut microbiome composition impacts outcomes in primary central nervous system lymphoma patients undergoing high-dose methotrexate-based chemotherapy. (A) Consensus heatmap showing microbiome community subtypes (CS1 and CS2). (B) Kaplan–Meier curves for progression-free and overall survival demonstrating a better outcome for CS1. (C) Hazard ratio estimates using progression-free survival (PFS). (D) Boxplot showing no difference in fecal diversity between community subtypes. (E) PCoA plot of fecal samples by CS group using Canberra distance. (F) Barplot of phylogenetic composition ordered by inverse Simpson diversity, with pie charts showing composition within each CS. (G) Boxplot showing enrichment of Parabacteroides distasonis in CS1. (H) Barplot of relative microbiome abundance showing independence from sampling before or after chemotherapy. — Gut microbiome composition impacts outcome in primary central nervous system lymphoma patients undergoing high-dose methotrexate-based chemotherapy treatment. (A) Consensus heatmap based on the 10 integrative clustering algorithms to refine the microbiome communities’ subtypes (CS1 and CS2). (B) Kaplan–Meier of progression-free survival (top) or overall survival (bottom) among patients belonging to each cluster. (C) Hazard-ratio estimates using the PFS. (D) Boxplot Fecal diversity between community subtypes (CS), calculated with inverse Simpson diversity index. P value calculated by two-sided Wilcoxon rank-sum test. The middle line is the median, the box limits represent the upper and lower quartiles, the whiskers note the 1.5× the iQr and the dots represent the individual data points. (E) Principal coordinate analysis of fecal samples by CS group using Canberra distance. P value and R of the distance between groups were calculated by ADONIS (n = 40; 12 vs. 28), while the beta dispersion within groups were calculated with a permutation test. (F) Barplot shows the phylogenic composition of fecal samples (n = 40) used in clustering. Samples are ordered by increasing inverse Simpson alpha diversity values and grouped by CS. Pie plots show the phylogenic composition within each CS group. (G) Boxplot showing that Parabacteroides distasonis is significantly enriched in CS1. P value calculated by two-sided Wilcoxon rank-sum test. The middle line is the median, the box limits represent the upper and lower quartiles, the whiskers note the 1.5× the iQr and the dots represent the individual data points. (H) Bar plot of relative abundance per patient showing that the distribution of microbiome values is independent of samples taken before or after the first chemotherapy. Calculated using Linear discriminant analysis Effect Size.

To better understand compositional differences between these microbial communities associated with clinical outcomes, we performed differential abundance analysis using the genus and the OTUs. The CS1 was especially enriched in Parabacteroides, whereas the CS2 in Prevotella (Figure 4F and Supplementary Figure 2C). The LEfSe⁴⁸ and MaAsLin3 analysis (compares relative and absolute abundance, respectively) using the chemotherapy sampling timing as a subclass, again revealed these bacteria as the top enriched in each CS (Supplementary Figure 2D and 2E). OTU level analysis revealed that within the Parabacteroides genus, Parabacteroides distasonis is particularly enriched in the CS1 (Figure 4G, Log₂-Fold Change [Log₂FC] = 2.52, P < .001, Supplementary Table 2) and remained significant even after adjusting for chemotherapy sampling using LEfSe and MaAsLin3 (Figure 4H and Supplementary Figure 2F). Finally, LEfSe and MaAsLin3 analysis on reconstructed metabolic pathways,^31,32 revealed an enrichment in diverse pathways like L-valine biosynthesis (VALSYN-PWY) and cis vaccenate biosynthesis (PWY 5973) in CS1; whereas, L-aspartate–aspargine biosynthesis (PWY ASPASN) was found enriched in CS2. Moreover, in CS1 most pathways were associated with Parabacteroides distasonis or Bacteroides fragilis and results were not linked to a lack of prevalence of these pathways in the CS2 group (Supplementary Figures 3 and 4). Thus, the gut microbiome composition impacts disease progression in PCNSL patients treated with HD-MTX-based CT, with Parabacteroides distasonis being among the top components of this effect.

Parabacteroides Distasonis’ High Content is Associated With an Improved Outcome to HD-MTX-based CT in PCNSL Patients

To further investigate the effect of Parabacteroides distasonis on treatment response, we stratified patients into either high versus low categories (based on the median or the second tertile), or high versus medium versus low categories (based on the first and second tertiles) using the relative abundance of this taxa in the gut microbiome. The second tertile was found to be the optimal cutoff and was consequently associated with the longest PFS and OS. Additionally, using the absolute abundance led to similar clinical associations. Based on these results, we decided to use the second tertile of the Parabacteroides distasonis’ relative abundance for subsequent analyses (Supplementary Figure 5). Patients presenting high Parabacteroides distasonis content had a significantly prolonged PFS in univariate (log rank P = .001, hazard ratio [HR] = 0.168, 95% CI = 0.575–0.049, 8.0 months vs. NR, 95% CI = 5.7–NR vs. 39–NR) and Cox proportional HR multivariate models (P = .002, HR = 0.140, 95% CI = 0.040–0.500, Figure 5A and Supplementary Figure 6A). In this group of patients, the effect on the OS was only significant at univariate level (P = .002, 37.2 months vs. NR, 95% CI = 11.4–NR vs. NR, Figure 5B). Next, we wondered whether or not the presence of this bacterium impacts immune cells’ infiltration into the CSF. Interestingly, we found higher percentage of CD3⁺CD8⁺ T cells in the Parabacteroides distasonis high group (Figure 5C) along with a positive linear correlation between this 2o variables (Supplementary Figure 6B); whereas, a trend of higher percentage of CD3⁺CD4⁺ T cells was observed in the Parabacteroides distasonis low group (Figure 5C), suggesting a link between host immune cell regulation and gut microbiome composition.

Association of Parabacteroides distasonis high content with improved outcomes in primary central nervous system lymphoma patients treated with high-dose methotrexate-based chemotherapy. (A) Kaplan–Meier curve and hazard ratio estimates for progression-free survival showing better outcome in patients with high P. distasonis content (≥66th percentile). (B) Kaplan–Meier curve for overall survival in the same group, with hazard ratio estimates not evaluable due to the absence of events in the second tertile. (C) Boxplot showing increased CD8+ T cell levels in the cerebrospinal fluid of the high P. distasonis group. — Parabacteroides distasonis high content is associated with an improved outcome to high-dose methotrexate-based chemotherapy in primary central nervous system lymphoma patients. (A) Kaplan–Meier (left) and Hazard-ratio estimates (right) of progression-free survival among patients with high content (≥Percentile 66) of Parabacteroides distasonis. (B) Kaplan–Meier of overall survival among patients with high content (≥Percentile 66) of Parabacteroides distasonis. Hazard ratio estimates on OS are non-evaluable due to the lack of events when grouping by the second tertile. (C) Boxplot showing that the Parabacteroides distasonis high group present higher CD8 + T cells in the CSF. P value calculated by two-sided Wilcoxon rank-sum test. The middle line is the median, the box limits represent the upper and lower quartiles, the whiskers note the 1.5× the iQr and the dots represent the individual data points.

Plasma Betaine–Valine Content is Associated With Parabacteroides Distasonis Content Along With Outcome After HD-MTX-based CT in PCNSL Patients

To investigate the metabolite changes associated with Parabacteroides distasonis content, we performed metabolomics on 90 plasma samples from HD-MTX-based CT-treated PCNSL patients of which 35 had metagenomic data. Patients belonging to the Parabacteroides distasonis high group showed significantly higher content of orotic, heptanoic, murideoxycholic, capric, sebacic, taurohyodeoxycholic, tauroursodeoxychlolic acids along with gamma-glutamyltryptophan and betaine–valine. Only ascorbic acid was significantly enriched in the low group (Figure 6A, Supplementary Table 3). To validate our clinical associations, we chose a metabolite based on its Spearman linear correlation with Parabacteroides distasonis content (n = 35) and its significant association with the OS and the PFS in the extended cohort (n = 90). While most of the enriched metabolites in the high group were linearly correlated with Parabacteroides distasonis (Supplementary Figure 6B), only the metabolite betaine/valine content (stratification also by the second tertile) was found associated with the PFS and OS using Cox proportional HR univariate models (Supplementary Figures 7 and 8). Interestingly, L-valine biosynthesis by Parabacteroides distasonis, using metagenomic’s reconstructed metabolic pathways, was previously found associated with the CS1 group (Supplementary Figure 3), reinforcing the idea that the presence of this metabolite reflects the presence of this bacterium. In this extended cohort of 90 patients, betaine–valine high content was found significantly associated with an increased PFS and OS using univariate models. Analysis with multivariate models adjusting by age as a continuous variable, KPS as binary, sex, and the arm of randomization showed significance for the PFS and a trend for the OS (Figure 6B and 6C and Supplementary Figures 9A and 9B); however, significance was reached for the OS when adjusting by age as a binary (by the median; Supplementary Figure 9C). To investigate the effect on the tumors, we used RNA-seq and WES sequencing data from our previous work.¹³ This included 9 tumors belonging to patients in the high betaine–valine group and 10 from the low betaine–valine group. Differential expression analysis using Microenvironment Cell Populations-counter (MCP-counter) cell population scores⁵⁵ and curated gene signatures¹³ revealed significantly higher macrophages as well as a trend towards higher immune checkpoint molecules, CD8 T cells, and memory-activated CD4 T cells (Figure 6D, Supplementary Table 4). Moreover, mutation frequency analysis showed no significant difference in the proportion of patients affected by the PCNSL hallmark mutations (Figure 6E, Supplementary Table 5). Altogether, these results suggest that Parabacteroides distasonis produces the betaine–valine metabolite and is associated with delaying disease progression by indirectly modulating CD8 T cell infiltration to the CSF and the tumor.

Association of plasma Betaine–Valine content with Parabacteroides distasonis levels and outcomes after high-dose methotrexate-based chemotherapy in primary central nervous system lymphoma patients. (A) Violin plots of the top ten differentially abundant plasma metabolites by P. distasonis content groups. (B) Kaplan–Meier curve and hazard ratio estimates for progression-free survival showing improved outcome in patients with high Betaine–Valine content (≥66th percentile). (C) Same as (B) for overall survival. (D) Violin plots of the top 5 MCP counter or gene signature scores from tumor RNA-seq data by Betaine–Valine content groups. (E) Barplot showing mutation rates of PCNSL driver genes by Betaine–Valine content, with colors representing genetic alteration types. — Plasma Betaine–Valine content is associated with Parabacteroides distasonis content along with outcome after high-dose methotrexate-based chemotherapy in primary central nervous system lymphoma patients. (A) Violin plots showing the top ten differentially abundant plasma metabolites according to the Parabacteroides distasonis content groups. (B) Kaplan–Meier (left) and Hazard-ratio estimates (right) of progression-free survival among patients with high content (≥Percentile 66) of Betaine–Valine metabolites. (C) Same as (B) but using the overall survival. (D) Violin plots showing the top 5 MCP counter or gene signatures scores from the tumors’ RNA-seq data according to Betaine–Valine content groups. P value calculated by two-sided Wilcoxon rank-sum test (A and D panels). The middle line is the median, the box limits represent the upper and lower quartiles, the whiskers note the 1.5× the iQr and the dots represent the individual data points. (E) Barplot showing the mutation rate of PCNSL drivers’ genes (inferred from WES data) according to Betaine–Valine content groups. Colors indicate the type of genetic alteration.

Discussion

To the best of our knowledge, this is the first study demonstrating a link between the gut microbiome composition and the outcome after HD-MTX-based CT treatment in PCNSL patients. Here, we report that gut microbial communities affect response to HD-MTX-based CT, more strongly on disease progression. Further inspection led us to discover Parabacteroides distasonis content as the main responsible, probably acting through the gut-brain axis, since we found higher CD8 T cell infiltration to the CSF in patients enriched with this bacterium. Moreover, from an analysis aimed to detect confounding factors, we found that antifungals, but not antibacterials or short-term HD-MTX-based CT treatment before sampling, disrupt the gut microbiome. Having proven that the metabolite betaine–valine is associated with Parabacteroides distasonis content, we used this metabolite in an extended cohort and validated our clinical associations. Finally, we showed that patients with high betaine–valine content, hence high Parabacteroides distasonis content, have an increased tumor infiltration of macrophages, immune checkpoint molecules, and CD8 T cells but not a distinct tumoral genetic profile.

In line with the only previous study investigating microbiome composition in 33 PCNSL patients, we found high abundances of Bacteroides, Firmicutes, Verrucomicrobia, Uroviricota, and Escherichia.²⁶ Gut microbiome associations with treatment response have been described in different cancer settings outside PCNSL.^{17–21,23–25} Smith et al. identified species within the class Clostridia that were associated with day 100 complete response to anti-CD29 chimeric antigen receptor (CAR) T-cell therapy in B-cell malignancies.¹⁹ In another study, Ruminococcaceae/Faecalibacterium and Akkermansia muciniphila high abundance was associated with melanoma responders patients treated with anti-PD1.¹⁸ In our study we found that HD-MTX-based CT treatment depletes Akkermansia muciniphila, suggesting that combination therapy with α-PD-1 should be given at early times rather than later. This Akkermansia muciniphila depletion due to MTX intake goes in line with a previous study.²¹ Our results do not support a Parabacteroides distasonis decrease due to HD-MTX-based CT, however we cannot rule out a long-term effect on this bacterium due to treatment.

Studies of Parabacteroides distasonis abundance as a protective factor have been reported in colorectal cancer through IL-10, TGF-β, and MyD88 regulation.^56,57 In line with our results, a recent study in bladder cancer showed that combination therapy of Parabacteroides distasonis + α-PD-1 significantly delayed tumor growth and increased the intratumoral densities of both CD4 + T and CD8 + T cells.⁵⁸ However, Routy et al. found Parabacteroides distasonis enriched in anti-PD1 treated non-responders patients with lung, renal, and urothelial tumors.²⁵ Betaine can prevent abnormal gene expression in cells by modulating DNA methylation and has been proven to have anti-inflammatory effects; however, its effect in cancer is not well elucidated.^59,60 In addition, Chen et al. showed that betaine consumption in mice led to enrichment of the taxa Parabacteroides, reinforcing a potential link between both.^61(p4)

Our study presents some limitations. We acknowledge the existence of various bias factors that could potentially impact the gut microbiome such as the diet, age inclusion criteria, concomitant medications (besides antibiotics), exercise, etc. which are impractical to eliminate. However, it is important to recognize that PCNSL predominantly affects elderly patients. Moreover, our conclusions could be further supported with an increased number of patients and/or a validation cohort with available gut microbiome and clinical data; however, this is at the moment difficult given the rarity of the disease. Finally, additional data is needed to elucidate the exact mechanism between the interplay of Parabacteroides distasonis, the betaine–valine metabolite, the immune system gut-brain regulation, and the HD-MTX-based CT response in PCNSL patients.

In conclusion, the gut microbiome, especially Parabacteroides distasonis, is likely to be a key modulator of HD-MTX-based CT response through distinct metabolites’ production, including betaine–valine, which ultimately leads to higher CD8 T cell infiltration to the CSF and the tumor. Further studies would be of paramount importance to validate our findings in this small cohort, to unravel the molecular mechanisms behind, and to provide further evidence that could support the rationale of using prebiotics or probiotics to increase this bacterium abundance prior to HD-MTX-based CT therapy initiation in PCNSL patients.

Supplementary Material

noaf059_suppl_Supplementary_Figures_S1-S9

noaf059_suppl_supplementary_figures_s1-s9.docx^{(1.6MB, docx)}

noaf059_suppl_Supplementary_Tables_S1-S5

noaf059_suppl_supplementary_tables_s1-s5.docx^{(62.9KB, docx)}

Acknowledgments

We thank Yannick Marie and Emeline Mundwiller (Paris Brain Institute), and Lucile Armenoult for the help with the microbiome DNA sequencing libraries’ preparation; Sylvère Durand (Gustave Roussy Cancer Campus, Plateforme de métabolomique) for the help with the plasma metabolomic data production; DRIDI-ALOULOU Amel (ONCONEUROTEK, Paris Brain Institute) for the plasma sample selection and preparation; and all the clinicians, pathologists, researchers, and patients associated with the study.

Contributor Information

Isaias Hernández-Verdin, Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Eva Kirasic, Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Karima Mokhtari, Department of Neuropathology, Groupe Hospitalier Pitié Salpêtrière, APHP, Paris, France; Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Noemie Barillot, Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Lucas Rincón de la Rosa, Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Elise Sourdeau, Department of Hematology, APHP, Hôpital Pitié-Salpêtrière and Sorbonne University, Paris, France.

Yahse Abada, Department of Neuroradiology, Groupe Hospitalier Pitié Salpêtrière, APHP, Paris, France.

Magali Le Tarff-Tavernier, Department of Hematology, APHP, Hôpital Pitié-Salpêtrière and Sorbonne University, Paris, France.

Lucia Nichelli, Department of Neuroradiology, Groupe Hospitalier Pitié Salpêtrière, APHP, Paris, France.

Laura Rozenblum, Sorbonne Université, AP-HP, Hôpitaux Universitaires Pitié-Salpêtrière Charles Foix, Service de Médecine Nucléaire and LIB, INSERM U1146, Paris, France.

Aurélie Kas, Sorbonne Université, AP-HP, Hôpitaux Universitaires Pitié-Salpêtrière Charles Foix, Service de Médecine Nucléaire and LIB, INSERM U1146, Paris, France.

Bertrand Mathon, Department of Neurosurgery, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France.

Sylvain Choquet, Department of Hematology, APHP, Hôpital Pitié-Salpêtrière and Sorbonne University, Paris, France.

Caroline Houillier, Department of Neurology-2, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France.

Khê Hoang-Xuan, Department of Neurology-2, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France; Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Agusti Alentorn, Department of Neurology-2, Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Pitié-Salpêtrière-Charles Foix, Paris, France; Institut du Cerveau-Paris Brain Institute-ICM, Inserm, Sorbonne Université, CNRS, Paris, France.

Funding

This work was supported by a grant from Investissements d’avenir and by the grant INCa-DGOS-Inserm_12560 of the SiRIC CURAMUS [grant number INCa-DGOS-Inserm_12560], the program “investissements d’avenir” [grant number ANR-10-IAIHU-06], PRT-K/INCa grant LOC-model reference 2017-1-RT-04, BETPSY project, overseen by the French National Research Agency, as part of the second “Investissements d’Avenir” program [grant number ANR-18-RHUS-0012], Foundation RAM active investments, ARTC foundation (no grant number), ANR JCJC grant LOCimm (ANR-23-CE17-0027) and Cancéropôle IdF Emergence 2024 grant.

Conflict of interest statement

L.N. declares payments from Olea Medical and Servier that are unrelated to the present manuscript. The remaining authors have no declaration of interests.

Authorship statement

I.H.V., and A.A. conceptualized the study, developed the methodology, and performed the visualization. E.K., and Y.A. performed the experiments. I.H.V., K.H.X. and A.A. performed the study investigation. I.H.V., E.K., N.B., L.R., and A.A. performed the formal analysis. K.M., E. S., M.T.T., S.C., C.H., and A.A. obtained the resources. I.H.V and A.A. performed the validation and wrote the original draft. All of the authors reviewed and edited the manuscript. A.A. obtained the funding and supervised the study.

Data availability

Microbiome data for this study have been deposited in the European Nucleotide Archive (ENA) at EMBL-EBI under accession number PRJEB78503 (https://www.ebi.ac.uk/ena/browser/view/PRJEB78503). RNA-seq and WES sequencing data used in this study was previously published¹³ and can be accessible at the European Genome-phenome Archive (EGAD00001008706; http://www.ebi.ac.uk/ega/). Metabolomic, and FC is available from the corresponding author upon reasonable request. No custom software code was developed for this study but specific code for reproducing the main figures is available at https://github.com/iS4i4S/PCNSLMicrobiome.

Consent for publication

Not applicable.

References

1. Deckert M, Batchelor TT, Ferry JA, et al. Primary diffuse large B cell lymphoma of the CNS. In: WHO Classification of Tumours Editorial Board, ed. World Health Organization Classification of Tumours of the Central Nervous System. 5th ed. Lyon, France: International Agency for Research on Cancer; 2021:351–355. [Google Scholar]
2. Eloranta S, Brånvall E, Celsing F, et al. Increasing incidence of primary central nervous system lymphoma but no improvement in survival in Sweden 2000-2013. Eur J Haematol. 2018;100(1):61–68. [DOI] [PubMed] [Google Scholar]
3. Farrall AL, Smith JR.. Changing incidence and survival of Primary Central Nervous System Lymphoma in Australia: A 33-year national population-based study. Cancers. 2021;13(3):403. [DOI] [PMC free article] [PubMed] [Google Scholar]
4. Haldorsen IS, Krossnes BK, Aarseth JH, et al. Increasing incidence and continued dismal outcome of primary central nervous system lymphoma in Norway 1989–2003: Time trends in a 15-year national survey. Cancer. 2007;110(8):1803–1814. [DOI] [PubMed] [Google Scholar]
5. Makino K, Nakamura H, Hide T, et al. Prognostic impact of completion of initial high-dose methotrexate therapy on primary central nervous system lymphoma: A single institution experience. Int J Clin Oncol. 2015;20(1):29–34. [DOI] [PubMed] [Google Scholar]
6. Mendez JS, Ostrom QT, Gittleman H, et al. The elderly left behind—changes in survival trends of primary central nervous system lymphoma over the past 4 decades. Neuro Oncol. 2018;20(5):687–694. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Houillier C, Soussain C, Ghesquières H, et al. Management and outcome of primary CNS lymphoma in the modern era: An LOC network study. Neurology. 2020;94(10):e1027–e1039. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Ferreri AJM, Calimeri T, Cwynarski K, et al. Primary central nervous system lymphoma. . Nat Rev Dis Primers. 2023;9(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Zhou Y, Liu W, Xu Z, et al. Analysis of genomic alteration in primary central nervous system lymphoma and the expression of some related genes. Neoplasia. 2018;20(10):1059–1069. [DOI] [PMC free article] [PubMed] [Google Scholar]
10. Radke J, Ishaque N, Koll R, et al. ; ICGC MMML-Seq Consortium. The genomic and transcriptional landscape of primary central nervous system lymphoma. Nat Commun. 2022;13(1):2558. [DOI] [PMC free article] [PubMed] [Google Scholar]
11. Wright GW, Huang DW, Phelan JD, et al. A probabilistic classification tool for genetic subtypes of diffuse large B cell lymphoma with therapeutic implications. Cancer Cell. 2020;37(4):551–568.e14. [DOI] [PMC free article] [PubMed] [Google Scholar]
12. Schmitz R, Wright GW, Huang DW, et al. Genetics and pathogenesis of diffuse large B-cell lymphoma. N Engl J Med. 2018;378(15):1396–1407. [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Hernández-Verdin I, Kirasic E, Wienand K, et al. Molecular and clinical diversity in primary central nervous system lymphoma. Ann Oncol. 2023;34(2):186–199. [DOI] [PubMed] [Google Scholar]
14. Hernández-Verdin I, Kirasic E, Wienand K, et al. Molecular and clinical diversity in primary central nervous system lymphoma. Ann Oncol. 2022;34(2):S0923753422047329. [DOI] [PubMed] [Google Scholar]
15. Alame M, Cornillot E, Cacheux V, et al. The immune contexture of primary central nervous system diffuse large B cell lymphoma associates with patient survival and specific cell signaling. Theranostics. 2021;11(8):3565–3579. [DOI] [PMC free article] [PubMed] [Google Scholar]
16. Marcelis L, Antoranz A, Delsupehe AM, et al. In-depth characterization of the tumor microenvironment in central nervous system lymphoma reveals implications for immune-checkpoint therapy. Cancer Immunol Immunother. 2020;69(9):1751–1766. [DOI] [PMC free article] [PubMed] [Google Scholar]
17. Helmink BA, Khan MAW, Hermann A, Gopalakrishnan V, Wargo JA.. The microbiome, cancer, and cancer therapy. Nat Med. 2019;25(3):377–388. [DOI] [PubMed] [Google Scholar]
18. Gopalakrishnan V, Spencer CN, Nezi L, et al. Gut microbiome modulates response to anti–PD-1 immunotherapy in melanoma patients. Science. 2018;359(6371):97–103. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Smith M, Dai A, Ghilardi G, et al. Gut microbiome correlates of response and toxicity following anti-CD19 CAR T cell therapy. Nat Med. 2022;28(4):713–723. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Hu Y, Li J, Ni F, et al. CAR-T cell therapy-related cytokine release syndrome and therapeutic response is modulated by the gut microbiome in hematologic malignancies. Nat Commun. 2022;13(1):5313. [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Nayak RR, Alexander M, Deshpande I, et al. Methotrexate impacts conserved pathways in diverse human gut bacteria leading to decreased host immune activation. Cell Host Microbe. 2021;29(3):362–377.e11. [DOI] [PMC free article] [PubMed] [Google Scholar]
22. Hanahan D. Hallmarks of cancer: New dimensions. Cancer Discov. 2022;12(1):31–46. [DOI] [PubMed] [Google Scholar]
23. Lin Z, Mao D, Jin C, et al. The gut microbiota correlate with the disease characteristics and immune status of patients with untreated diffuse large B-cell lymphoma. Front Immunol. 2023;14:1105293. [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Yuan L, Wang W, Zhang W, et al. Gut microbiota in untreated diffuse large B cell lymphoma patients. Front Microbiol. 2021;12:646361. [DOI] [PMC free article] [PubMed] [Google Scholar]
25. Routy B, Le Chatelier E, Derosa L, et al. Gut microbiome influences efficacy of PD-1–based immunotherapy against epithelial tumors. Science. 2018;359(6371):91–97. [DOI] [PubMed] [Google Scholar]
26. Kang Z, Zhang R, Wang C, et al. Integrated analysis of fecal metagenomics and metabolomics reveals the role of gut microbiota in the pathogenesis of primary central nervous system lymphoma. Research Square. 2023. doi: https://doi.org/ 10.21203/rs.3.rs-3051515/v1 [DOI]
27. Langmead B, Salzberg SL.. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9(4):357–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Wood DE, Lu J, Langmead B.. Improved metagenomic analysis with Kraken 2. Genome Biol. 2019;20(1):257. [DOI] [PMC free article] [PubMed] [Google Scholar]
29. Lu J, Rincon N, Wood DE, et al. Metagenome analysis using the Kraken software suite. Nat Protoc. 2022;17(12):2815–2839. [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Lu J, Breitwieser FP, Thielen P, Salzberg SL.. Bracken: Estimating species abundance in metagenomics data. PeerJ Comput Sci. 2017;3:e104. [DOI] [PMC free article] [PubMed] [Google Scholar]
31. Beghini F, McIver LJ, Blanco-Míguez A, et al. Integrating taxonomic, functional, and strain-level profiling of diverse microbial communities with bioBakery 3. Elife. 2021;10:e65088. [DOI] [PMC free article] [PubMed] [Google Scholar]
32. Krieger CJ. MetaCyc: A multiorganism database of metabolic pathways and enzymes. Nucleic Acids Res. 2004;32(90001):438D–4442. [DOI] [PMC free article] [PubMed] [Google Scholar]
33. McMurdie PJ, Holmes S.. An R package for reproducible interactive analysis and graphics of microbiome census data. Watson M, ed. PLoS One. 2013;8(4):e61217. [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Oksanen J, Simpson GL, Blanchet FG, et al. Vegan: Community Ecology Package.; 2022. https://CRAN.R-project.org/package=vegan [Google Scholar]
35. Love MI, Huber W, Anders S.. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Mo Q, Shen R, Guo C, et al. A fully Bayesian latent variable model for integrative clustering analysis of multi-type omics data. Biostatistics. 2018;19(1):71–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Meng C, Helm D, Frejno M, Kuster B.. moCluster: Identifying joint patterns across multiple omics data sets. J Proteome Res. 2016;15(3):755–765. [DOI] [PubMed] [Google Scholar]
38. Ramazzotti D, Lal A, Wang B, Batzoglou S, Sidow A.. Multi-omic tumor data reveal diversity of molecular mechanisms that correlate with survival. Nat Commun. 2018;9(1):4453. [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Chalise P, Fridley BL.. Integrative clustering of multi-level ‘omic data based on non-negative matrix factorization algorithm. Peddada SD, ed. PLoS One. 2017;12(5):e0176278. [DOI] [PMC free article] [PubMed] [Google Scholar]
40. Monti S. [No title found]. Mach Learn. 2003;52(1/2):91–118. [Google Scholar]
41. Hoadley KA, Yau C, Wolf DM, et al. ; Cancer Genome Atlas Research Network. Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin. Cell. 2014;158(4):929–944. [DOI] [PMC free article] [PubMed] [Google Scholar]
42. Rappoport N, Shamir R.. NEMO: Cancer subtyping by integration of partial multi-omic data. Schwartz R, ed. Bioinformatics. 2019;35(18):3348–3356. [DOI] [PMC free article] [PubMed] [Google Scholar]
43. Nguyen H, Shrestha S, Draghici S, Nguyen T.. PINSPlus: A tool for tumor subtype discovery in integrated genomic data. Berger B, ed. Bioinformatics. 2019;35(16):2843–2846. [DOI] [PubMed] [Google Scholar]
44. Wang B, Mezlini AM, Demir F, et al. Similarity network fusion for aggregating data types on a genomic scale. Nat Methods. 2014;11(3):333–337. [DOI] [PubMed] [Google Scholar]
45. Wu D, Wang D, Zhang MQ, Gu J.. Fast dimension reduction and integrative clustering of multi-omics data using low-rank approximation: Application to cancer molecular classification. BMC Genomics. 2015;16(1):1022. [DOI] [PMC free article] [PubMed] [Google Scholar]
46. Lu X, Meng J, Zhou Y, Jiang L, Yan F.. MOVICS: An R package for multi-omics integration and visualization in cancer subtyping. Martelli PL, ed. Bioinformatics. 2021;36(22-23):5539–5541. [DOI] [PubMed] [Google Scholar]
47. Rousseeuw PJ. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J Comput Appl Math. 1987;20(1):53–65. [Google Scholar]
48. Segata N, Izard J, Waldron L, et al. Metagenomic biomarker discovery and explanation. Genome Biol. 2011;12(6):R60. [DOI] [PMC free article] [PubMed] [Google Scholar]
49. Mallick H, Rahnavard A, McIver LJ, et al. Multivariable association discovery in population-scale meta-omics studies. Coelho LP, ed. PLoS Comput Biol. 2021;17(11):e1009442. [DOI] [PMC free article] [PubMed] [Google Scholar]
50. Grajeda-Iglesias C, Durand S, Daillère R, et al. Oral administration of Akkermansia muciniphila elevates systemic antiaging and anticancer metabolites. Aging (Milano). 2021;13(5):6375–6405. [DOI] [PMC free article] [PubMed] [Google Scholar]
51. Alves Costa Silva C, Piccinno G, Suissa D, et al. Influence of microbiota-associated metabolic reprogramming on clinical outcome in patients with melanoma from the randomized adjuvant dendritic cell-based MIND-DC trial. Nat Commun. 2024;15(1):1633. [DOI] [PMC free article] [PubMed] [Google Scholar]
52. Durand S, Grajeda-Iglesias C, Aprahamian F, Nirmalathasan N, Kepp O, Kroemer G.. The intracellular metabolome of starving cells. In: Wilson L, Matsudaira P, eds. Methods in Cell Biology. Vol 164. Amsterdam, Netherlands: Elsevier; 2021:137–156. doi: https://doi.org/ 10.1016/bs.mcb.2021.04.001 [DOI] [PubMed] [Google Scholar]
53. Abdellatif M, Trummer-Herbst V, Koser F, et al. Nicotinamide for the treatment of heart failure with preserved ejection fraction. Sci Transl Med. 2021;13(580):eabd7064. [DOI] [PMC free article] [PubMed] [Google Scholar]
54. Armand M, Costopoulos M, Osman J, et al. Optimization of CSF biological investigations for CNS lymphoma diagnosis. Am J Hematol. 2019;94(10):1123–1131. [DOI] [PubMed] [Google Scholar]
55. Becht E, Giraldo NA, Lacroix L, et al. Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol. 2016;17(1):218. [DOI] [PMC free article] [PubMed] [Google Scholar]
56. Ezeji JC, Sarikonda DK, Hopperton A, et al. Parabacteroides distasonis: Intriguing aerotolerant gut anaerobe with emerging antimicrobial resistance and pathogenic and probiotic roles in human health. Gut Microbes. 2021;13(1):1922241. [DOI] [PMC free article] [PubMed] [Google Scholar]
57. Koh GY, Kane AV, Wu X, Crott JW.. Parabacteroides distasonis attenuates tumorigenesis, modulates inflammatory markers and promotes intestinal barrier integrity in azoxymethane-treated A/J mice. Carcinogenesis. 2020;41(7):909–917. [DOI] [PubMed] [Google Scholar]
58. Wang B, Qiu Y, Xie M, et al. Gut microbiota Parabacteroides distasonis enchances the efficacy of immunotherapy for bladder cancer by activating anti-tumor immune responses. BMC Microbiol. 2024;24(1):237. [DOI] [PMC free article] [PubMed] [Google Scholar]
59. Zhou R, Chen XL, Zhou Z, et al. Higher dietary intakes of choline and betaine are associated with a lower risk of primary liver cancer: A case-control study. Sci Rep. 2017;7(1):679. [DOI] [PMC free article] [PubMed] [Google Scholar]
60. Zhao G, He F, Wu C, et al. Betaine in inflammation: Mechanistic aspects and applications. Front Immunol. 2018;9:1070. [DOI] [PMC free article] [PubMed] [Google Scholar]
61. Chen Q, Wang Y, Jiao F, et al. Betaine inhibits Toll-like receptor 4 responses and restores intestinal microbiota in acute liver failure mice. Sci Rep. 2020;10(1):21850. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

noaf059_suppl_Supplementary_Figures_S1-S9

noaf059_suppl_supplementary_figures_s1-s9.docx^{(1.6MB, docx)}

noaf059_suppl_Supplementary_Tables_S1-S5

noaf059_suppl_supplementary_tables_s1-s5.docx^{(62.9KB, docx)}

Data Availability Statement

[CIT0001] 1. Deckert M, Batchelor TT, Ferry JA, et al. Primary diffuse large B cell lymphoma of the CNS. In: WHO Classification of Tumours Editorial Board, ed. World Health Organization Classification of Tumours of the Central Nervous System. 5th ed. Lyon, France: International Agency for Research on Cancer; 2021:351–355. [Google Scholar]

[CIT0002] 2. Eloranta S, Brånvall E, Celsing F, et al. Increasing incidence of primary central nervous system lymphoma but no improvement in survival in Sweden 2000-2013. Eur J Haematol. 2018;100(1):61–68. [DOI] [PubMed] [Google Scholar]

[CIT0003] 3. Farrall AL, Smith JR.. Changing incidence and survival of Primary Central Nervous System Lymphoma in Australia: A 33-year national population-based study. Cancers. 2021;13(3):403. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0004] 4. Haldorsen IS, Krossnes BK, Aarseth JH, et al. Increasing incidence and continued dismal outcome of primary central nervous system lymphoma in Norway 1989–2003: Time trends in a 15-year national survey. Cancer. 2007;110(8):1803–1814. [DOI] [PubMed] [Google Scholar]

[CIT0005] 5. Makino K, Nakamura H, Hide T, et al. Prognostic impact of completion of initial high-dose methotrexate therapy on primary central nervous system lymphoma: A single institution experience. Int J Clin Oncol. 2015;20(1):29–34. [DOI] [PubMed] [Google Scholar]

[CIT0006] 6. Mendez JS, Ostrom QT, Gittleman H, et al. The elderly left behind—changes in survival trends of primary central nervous system lymphoma over the past 4 decades. Neuro Oncol. 2018;20(5):687–694. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0007] 7. Houillier C, Soussain C, Ghesquières H, et al. Management and outcome of primary CNS lymphoma in the modern era: An LOC network study. Neurology. 2020;94(10):e1027–e1039. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0008] 8. Ferreri AJM, Calimeri T, Cwynarski K, et al. Primary central nervous system lymphoma. . Nat Rev Dis Primers. 2023;9(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0009] 9. Zhou Y, Liu W, Xu Z, et al. Analysis of genomic alteration in primary central nervous system lymphoma and the expression of some related genes. Neoplasia. 2018;20(10):1059–1069. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0010] 10. Radke J, Ishaque N, Koll R, et al. ; ICGC MMML-Seq Consortium. The genomic and transcriptional landscape of primary central nervous system lymphoma. Nat Commun. 2022;13(1):2558. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0011] 11. Wright GW, Huang DW, Phelan JD, et al. A probabilistic classification tool for genetic subtypes of diffuse large B cell lymphoma with therapeutic implications. Cancer Cell. 2020;37(4):551–568.e14. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0012] 12. Schmitz R, Wright GW, Huang DW, et al. Genetics and pathogenesis of diffuse large B-cell lymphoma. N Engl J Med. 2018;378(15):1396–1407. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0013] 13. Hernández-Verdin I, Kirasic E, Wienand K, et al. Molecular and clinical diversity in primary central nervous system lymphoma. Ann Oncol. 2023;34(2):186–199. [DOI] [PubMed] [Google Scholar]

[CIT0014] 14. Hernández-Verdin I, Kirasic E, Wienand K, et al. Molecular and clinical diversity in primary central nervous system lymphoma. Ann Oncol. 2022;34(2):S0923753422047329. [DOI] [PubMed] [Google Scholar]

[CIT0015] 15. Alame M, Cornillot E, Cacheux V, et al. The immune contexture of primary central nervous system diffuse large B cell lymphoma associates with patient survival and specific cell signaling. Theranostics. 2021;11(8):3565–3579. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0016] 16. Marcelis L, Antoranz A, Delsupehe AM, et al. In-depth characterization of the tumor microenvironment in central nervous system lymphoma reveals implications for immune-checkpoint therapy. Cancer Immunol Immunother. 2020;69(9):1751–1766. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0017] 17. Helmink BA, Khan MAW, Hermann A, Gopalakrishnan V, Wargo JA.. The microbiome, cancer, and cancer therapy. Nat Med. 2019;25(3):377–388. [DOI] [PubMed] [Google Scholar]

[CIT0018] 18. Gopalakrishnan V, Spencer CN, Nezi L, et al. Gut microbiome modulates response to anti–PD-1 immunotherapy in melanoma patients. Science. 2018;359(6371):97–103. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0019] 19. Smith M, Dai A, Ghilardi G, et al. Gut microbiome correlates of response and toxicity following anti-CD19 CAR T cell therapy. Nat Med. 2022;28(4):713–723. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0020] 20. Hu Y, Li J, Ni F, et al. CAR-T cell therapy-related cytokine release syndrome and therapeutic response is modulated by the gut microbiome in hematologic malignancies. Nat Commun. 2022;13(1):5313. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0021] 21. Nayak RR, Alexander M, Deshpande I, et al. Methotrexate impacts conserved pathways in diverse human gut bacteria leading to decreased host immune activation. Cell Host Microbe. 2021;29(3):362–377.e11. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0022] 22. Hanahan D. Hallmarks of cancer: New dimensions. Cancer Discov. 2022;12(1):31–46. [DOI] [PubMed] [Google Scholar]

[CIT0023] 23. Lin Z, Mao D, Jin C, et al. The gut microbiota correlate with the disease characteristics and immune status of patients with untreated diffuse large B-cell lymphoma. Front Immunol. 2023;14:1105293. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0024] 24. Yuan L, Wang W, Zhang W, et al. Gut microbiota in untreated diffuse large B cell lymphoma patients. Front Microbiol. 2021;12:646361. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0025] 25. Routy B, Le Chatelier E, Derosa L, et al. Gut microbiome influences efficacy of PD-1–based immunotherapy against epithelial tumors. Science. 2018;359(6371):91–97. [DOI] [PubMed] [Google Scholar]

[CIT0026] 26. Kang Z, Zhang R, Wang C, et al. Integrated analysis of fecal metagenomics and metabolomics reveals the role of gut microbiota in the pathogenesis of primary central nervous system lymphoma. Research Square. 2023. doi: https://doi.org/ 10.21203/rs.3.rs-3051515/v1 [DOI]

[CIT0027] 27. Langmead B, Salzberg SL.. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9(4):357–359. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0028] 28. Wood DE, Lu J, Langmead B.. Improved metagenomic analysis with Kraken 2. Genome Biol. 2019;20(1):257. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0029] 29. Lu J, Rincon N, Wood DE, et al. Metagenome analysis using the Kraken software suite. Nat Protoc. 2022;17(12):2815–2839. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0030] 30. Lu J, Breitwieser FP, Thielen P, Salzberg SL.. Bracken: Estimating species abundance in metagenomics data. PeerJ Comput Sci. 2017;3:e104. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0031] 31. Beghini F, McIver LJ, Blanco-Míguez A, et al. Integrating taxonomic, functional, and strain-level profiling of diverse microbial communities with bioBakery 3. Elife. 2021;10:e65088. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0032] 32. Krieger CJ. MetaCyc: A multiorganism database of metabolic pathways and enzymes. Nucleic Acids Res. 2004;32(90001):438D–4442. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0033] 33. McMurdie PJ, Holmes S.. An R package for reproducible interactive analysis and graphics of microbiome census data. Watson M, ed. PLoS One. 2013;8(4):e61217. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0034] 34. Oksanen J, Simpson GL, Blanchet FG, et al. Vegan: Community Ecology Package.; 2022. https://CRAN.R-project.org/package=vegan [Google Scholar]

[CIT0035] 35. Love MI, Huber W, Anders S.. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0036] 36. Mo Q, Shen R, Guo C, et al. A fully Bayesian latent variable model for integrative clustering analysis of multi-type omics data. Biostatistics. 2018;19(1):71–86. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0037] 37. Meng C, Helm D, Frejno M, Kuster B.. moCluster: Identifying joint patterns across multiple omics data sets. J Proteome Res. 2016;15(3):755–765. [DOI] [PubMed] [Google Scholar]

[CIT0038] 38. Ramazzotti D, Lal A, Wang B, Batzoglou S, Sidow A.. Multi-omic tumor data reveal diversity of molecular mechanisms that correlate with survival. Nat Commun. 2018;9(1):4453. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0039] 39. Chalise P, Fridley BL.. Integrative clustering of multi-level ‘omic data based on non-negative matrix factorization algorithm. Peddada SD, ed. PLoS One. 2017;12(5):e0176278. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0040] 40. Monti S. [No title found]. Mach Learn. 2003;52(1/2):91–118. [Google Scholar]

[CIT0041] 41. Hoadley KA, Yau C, Wolf DM, et al. ; Cancer Genome Atlas Research Network. Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin. Cell. 2014;158(4):929–944. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0042] 42. Rappoport N, Shamir R.. NEMO: Cancer subtyping by integration of partial multi-omic data. Schwartz R, ed. Bioinformatics. 2019;35(18):3348–3356. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0043] 43. Nguyen H, Shrestha S, Draghici S, Nguyen T.. PINSPlus: A tool for tumor subtype discovery in integrated genomic data. Berger B, ed. Bioinformatics. 2019;35(16):2843–2846. [DOI] [PubMed] [Google Scholar]

[CIT0044] 44. Wang B, Mezlini AM, Demir F, et al. Similarity network fusion for aggregating data types on a genomic scale. Nat Methods. 2014;11(3):333–337. [DOI] [PubMed] [Google Scholar]

[CIT0045] 45. Wu D, Wang D, Zhang MQ, Gu J.. Fast dimension reduction and integrative clustering of multi-omics data using low-rank approximation: Application to cancer molecular classification. BMC Genomics. 2015;16(1):1022. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0046] 46. Lu X, Meng J, Zhou Y, Jiang L, Yan F.. MOVICS: An R package for multi-omics integration and visualization in cancer subtyping. Martelli PL, ed. Bioinformatics. 2021;36(22-23):5539–5541. [DOI] [PubMed] [Google Scholar]

[CIT0047] 47. Rousseeuw PJ. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J Comput Appl Math. 1987;20(1):53–65. [Google Scholar]

[CIT0048] 48. Segata N, Izard J, Waldron L, et al. Metagenomic biomarker discovery and explanation. Genome Biol. 2011;12(6):R60. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0049] 49. Mallick H, Rahnavard A, McIver LJ, et al. Multivariable association discovery in population-scale meta-omics studies. Coelho LP, ed. PLoS Comput Biol. 2021;17(11):e1009442. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0050] 50. Grajeda-Iglesias C, Durand S, Daillère R, et al. Oral administration of Akkermansia muciniphila elevates systemic antiaging and anticancer metabolites. Aging (Milano). 2021;13(5):6375–6405. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0051] 51. Alves Costa Silva C, Piccinno G, Suissa D, et al. Influence of microbiota-associated metabolic reprogramming on clinical outcome in patients with melanoma from the randomized adjuvant dendritic cell-based MIND-DC trial. Nat Commun. 2024;15(1):1633. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0052] 52. Durand S, Grajeda-Iglesias C, Aprahamian F, Nirmalathasan N, Kepp O, Kroemer G.. The intracellular metabolome of starving cells. In: Wilson L, Matsudaira P, eds. Methods in Cell Biology. Vol 164. Amsterdam, Netherlands: Elsevier; 2021:137–156. doi: https://doi.org/ 10.1016/bs.mcb.2021.04.001 [DOI] [PubMed] [Google Scholar]

[CIT0053] 53. Abdellatif M, Trummer-Herbst V, Koser F, et al. Nicotinamide for the treatment of heart failure with preserved ejection fraction. Sci Transl Med. 2021;13(580):eabd7064. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0054] 54. Armand M, Costopoulos M, Osman J, et al. Optimization of CSF biological investigations for CNS lymphoma diagnosis. Am J Hematol. 2019;94(10):1123–1131. [DOI] [PubMed] [Google Scholar]

[CIT0055] 55. Becht E, Giraldo NA, Lacroix L, et al. Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol. 2016;17(1):218. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0056] 56. Ezeji JC, Sarikonda DK, Hopperton A, et al. Parabacteroides distasonis: Intriguing aerotolerant gut anaerobe with emerging antimicrobial resistance and pathogenic and probiotic roles in human health. Gut Microbes. 2021;13(1):1922241. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0057] 57. Koh GY, Kane AV, Wu X, Crott JW.. Parabacteroides distasonis attenuates tumorigenesis, modulates inflammatory markers and promotes intestinal barrier integrity in azoxymethane-treated A/J mice. Carcinogenesis. 2020;41(7):909–917. [DOI] [PubMed] [Google Scholar]

[CIT0058] 58. Wang B, Qiu Y, Xie M, et al. Gut microbiota Parabacteroides distasonis enchances the efficacy of immunotherapy for bladder cancer by activating anti-tumor immune responses. BMC Microbiol. 2024;24(1):237. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0059] 59. Zhou R, Chen XL, Zhou Z, et al. Higher dietary intakes of choline and betaine are associated with a lower risk of primary liver cancer: A case-control study. Sci Rep. 2017;7(1):679. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0060] 60. Zhao G, He F, Wu C, et al. Betaine in inflammation: Mechanistic aspects and applications. Front Immunol. 2018;9:1070. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CIT0061] 61. Chen Q, Wang Y, Jiao F, et al. Betaine inhibits Toll-like receptor 4 responses and restores intestinal microbiota in acute liver failure mice. Sci Rep. 2020;10(1):21850. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Gut microbiome modulates the outcome in primary central nervous system lymphoma patients undergoing chemotherapy: An ancillary study from the BLOCAGE trial

Isaias Hernández-Verdin

Eva Kirasic

Karima Mokhtari

Noemie Barillot

Lucas Rincón de la Rosa

Elise Sourdeau

Yahse Abada

Magali Le Tarff-Tavernier

Lucia Nichelli

Laura Rozenblum

Aurélie Kas

Bertrand Mathon

Sylvain Choquet

Caroline Houillier

Khê Hoang-Xuan

Agusti Alentorn

Abstract

Background

Methods

Results

Conclusions

Key Points.

Importance of the Study.

Methods

Patients

Figure 1.

Fecal Microbiome Collection and Processing

Figure 2.

Metagenomic Shotgun Sequencing and Analysis

Metabolomics

Flow Cytometry on CSF Samples

Statistical Analyses

Ethics Statement

Results

Distribution of Gut Microbiota in PCNSL Patients

Antifungals Intake Before HD-MTX-based CT Treatment Affects Gut Microbiome in PCNSL Patients

HD-MTX-based CT Treatment Depletes Akkermansia Muciniphila and Bifidobacterium Dentium Gut Microbiome Content in PCNSL Patients

Figure 3.

Gut Microbiome Composition Impacts Outcome in PCNSL Patients Undergoing HD-MTX-based CT Treatment

Figure 4.

Parabacteroides Distasonis’ High Content is Associated With an Improved Outcome to HD-MTX-based CT in PCNSL Patients

Figure 5.

Plasma Betaine–Valine Content is Associated With Parabacteroides Distasonis Content Along With Outcome After HD-MTX-based CT in PCNSL Patients

Figure 6.

Discussion

Supplementary Material

Acknowledgments

Contributor Information

Funding

Conflict of interest statement

Authorship statement

Data availability

Consent for publication

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases