Sang Eun Yoon; Woorim Kang; Junhun Cho; Mauricio Chalita; Je Hee Lee; Dong-Wook Hyun; Hyun Kim; Seok Jin Kim; Won Seog Kim

doi:10.4143/crt.2024.675

Articles

Page Path: HOME > Cancer Res Treat > Volume 57(2); 2025 > Article

Original Article Hematologic malignancy Clinical Impact of Microbiome Characteristics in Treatment-Naïve Extranodal NK/T-Cell Lymphoma Patients: Sang Eun Yoon¹, Woorim Kang², Junhun Cho³, Mauricio Chalita², Je Hee Lee², Dong-Wook Hyun², Hyun Kim², Seok Jin Kim¹, Won Seog Kim^1,; Cancer Research and Treatment : Official Journal of Korean Cancer Association 2025;57(2):597-611.
DOI: https://doi.org/10.4143/crt.2024.675
Published online: August 16, 2024

¹Division of Hematology-Oncology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea

²CJ Bioscience Inc., Seoul, Korea

³Department of Pathology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea

Correspondence: Won Seog Kim, Division of Hematology-Oncology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, Korea
Tel: 82-2-3410-6548 E-mail: wskimsmc@skku.edu

*Sang Eun Yoon and Woorim Kang contributed equally to this work.

• Received: July 22, 2024 • Accepted: August 14, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

1,171 Views
109 Download

prev next

Full Article

Download PDF

Abstract

Purpose
Extranodal natural killer/T-cell lymphoma (ENKTL) predominantly manifests in East Asia and Latin America. Despite shared intrinsic factors, such as ethnic and genetic backgrounds, the progression of ENKTL can be influenced by extrinsic factors related to changing lifestyle patterns.
Materials and Methods
This study collected stool samples from newly diagnosed (ND)–ENKTL patients (n=40) and conducted whole genome shotgun sequencing.
Results
ND-ENKTL revealed reduced alpha diversity in ND-ENKTL compared to healthy controls (HCs) (p=0.008), with Enterobacteriaceae abundance significantly contributing to the beta diversity difference between ENKTL and HCs (p < 0.001). Functional analysis indicated upregulated aerobic metabolism and degradation of aromatic compounds in ND-ENKTL. Enterobacteriaceae were associated not only with clinical data explaining disease status (serum C-reactive protein, stage, prognosis index of natural killer cell lymphoma [PINK], and PINK-E) but also with clinical outcomes (early relapse and short progression-free survival). The relative abundance of Enterobacteriaceae at the family level was similar between ENKTL and diffuse large B-cell lymphoma (DLBCL) (p=0.140). However, the ENKTL exhibited a higher abundance of Escherichia, in contrast to the prevalence of Enterobacter and Citrobacter in DLBCL. Linear regression analysis demonstrated a significant association between Escherichia abundance and programmed cell death-ligand-1 (PD-L1) levels in tissue samples (p=0.025), whereas no correlation with PD-L1 was observed for Enterobacteriaceae at the family level (p=0.571).
Conclusion
ND-ENKTL exhibited an abundance of Enterobacteriaceae and a dominant presence of Escherichia. These microbial characteristics correlated with disease status, treatment outcomes, and PD-L1 expression, suggesting the potential of the ENKTL microbiome as a biomarker and cause of lymphomagenesis, which warrants further exploration.
Key words: Extranodal NK/T-cell lymphoma, Microbiome, Diversity, Abundance, Diffuse large B-cell lymphoma

Introduction

Epstein-Barr virus (EBV) establishes lifelong persistence within the human host through B-cell infection and residence within memory B cells. This phenomenon occurs predominantly in healthy individuals and is characterized by its asymptomatic nature and the absence of disease manifestation [1]. Among EBV-associated diseases, extranodal natural killer/T-cell lymphoma (ENKTL), an aggressive subtype of extranodal non-Hodgkin lymphoma, mostly occurs in extranodal sites such as the nose, nasopharynx, and oropharyngeal regions with extensive necrosis, inflammation, and vascular damage [2]. However, insufficient evidence exists on the triggers of ENKTL development and the reason for the predominant localization of the disease within the upper respiratory tract, the initial site of EBV infection.

According to genome-wide association studies of patients with ENKTL in Asia, two intrinsic genetic factors, IL18RAP on chromosome 2 and HLA-DRB1 on chromosome 6, were reported to affect lymphomagenesis [3,4]. The researchers suggested that the development of ENKTL is through immune regulation via the IL18RAP axis and antigen presentation involving HLA-DRB1. Interestingly, the incidence of ENKTL in British Columbia is lower in Chinese immigrants than in the national Chinese population, even though they share the same ethnicity. This underscores that despite sharing the same ethnic background, the progression of ENKTL is influenced by two distinct factors: intrinsic genetic susceptibility and extrinsic lifestyle factors. The evolution of the gastrointestinal microbiome depends on geography, ethnicity, or life-specific variations worldwide [5-7]. Moreover, it is reported that intestinal microbiota core composition undergoes progressive changes in response to changes in host survival strategies. Thus, it could be assumed that ENKTL prevalence, limited to specific regions, might be related to differences in the microbial community according to geography, ethnicity, or life-specific variations.

Previously, a comparison of the nasal cavity resident microbiome was undertaken between 46 ENKTL patients and 24 matched healthy controls (HCs) using 16S rRNA sequencing. Based on this study, the genus Staphylococcus was the most abundant in the ENKTL-related microbiome [8]. According to another study, involving the shotgun sequencing of 30 ENKTL patients, those with a higher relative abundance of Streptococcus parasanguinis or Romboutsia timonensis showed a significant correlation with advanced-stage disease and worse survival outcomes [9]. However, the two studies were analyzed with relatively small sample numbers and different sample types. Therefore, there is still limited evidence to understand the pathogenesis of the microbiome or to propose a promising biomarker in clinical practice.

Therefore, we conducted a stool analysis to determine the microbiome characteristics of newly diagnosed (ND)–ENKTL. Additionally, we analyzed the gut microbiome of ND–diffuse large B-cell lymphoma (DLBCL) patients using whole genome shotgun sequencing in advance. The abundance of the Enterobacteriaceae family belonging to the Proteobacteria phylum was higher in ND-DLBCL patients than in HCs. Moreover, the ND-DLBCL–specific microbiome could influence immunotherapy efficacies and incidence of toxicities [10]. Based on previous ND-DLBCL data, a comparison of microbiome composition among healthy individuals, ND-DLBCL, and ND-ENKTL was performed to identify factors that could influence ENKTL development and outcomes.

Materials and Methods

1. Study design and data collection

We have gathered and analyzed microbial data from 40 feces samples obtained from 40 ND-ENKTL patients from 2020 to 2022. Fecal samples from ENKTL patients were collected using a stool sampling kit (CJ Bioscience, Inc.) designed explicitly for microbial community analysis using shotgun sequencing. Pre-analyzed gut microbiome data were used from HC or ND-DLBCL groups to compare gut microbiome specificities [10]. We gathered clinical information at diagnosis, including sex, age, complete blood count, C-reactive protein (CRP), EBV titer (copies/mL), distant lymph node involvement, extranodal involvement, bone marrow involvement, Ann Arbor stage, prognosis index of natural killer cell lymphoma (PINK), and PINK with EBV DNA (PINK-E) [11]. The limit of EBV DNA detection in our institute was 110 copies/mL. Thus, a classification of EBV negative was determined when less than 110 copies/mL of EBV DNA was detected. We collected programmed cell death-ligand-1 (PD-L1) expression data using a PD-L1 IHC 22C3 pharmDx kit and the Dako ASL48 platform (Ventana). PD-L1 positivity was defined as membranous PD-L1 expression by more than 10% of tumor cells, regardless of staining intensity. We previously evaluated the tumor immune microenvironment (TIME) of ENKTL based on the immunohistochemical (IHC) markers FoxP3, PD-L1, and CD68 [12]. Following IHC findings, we classified four distinct TIME subtypes: immune tolerant, immune evasive A, immune evasive B, and immune silenced.

In addition, we assessed the historical treatment records of the patients, including concurrent chemoradiotherapy, systemic chemotherapy, autologous stem cell transplantation, and allogeneic stem cell transplantation. Treatment outcomes were assessed by 18F-fluorodeoxyglucose positron emission tomography/computed tomography (FDG PET/CT) and EBV DNA presence/absence. FDG PET/CT findings were used to determine the deauville scale (DS) based on the International Working Group response criteria [13]. Depending on the DS, the overall response rate was defined as the summation of complete remission and partial response. The final update of survival data was performed in July 2023.

2. Whole genome shotgun metagenomic sequencing

Libraries for whole genome shotgun metagenomic sequencing were created using an NEBNext Ultra II DNA Library Prep Kit and NEBNext Multiplex Oligos for Illumina from New England Biolabs, following the supplier’s instructions. The size and concentration of the DNA in the final library were verified via a Bioanalyzer system from Agilent Technologies, and sequenced on the Illumina NextSeq 1000 platform (2×250 bp read length) at CJ Bioscience Inc.

3. Taxonomic profiling of shotgun metagenomics data

A Kraken2 database, incorporating bacterial and archaeal species from the EzBioCloud database, was created. Ninety-two essential genes were identified using the UBCG pipeline for subsequent analysis. We developed a taxonomic framework compatible with Kraken2, based on the taxonomy in EzBioCloud, converting core gene sequences into FASTA files using unique numerical identifiers. The database construction used the Kraken2-build command with a k-size of 35 and standard settings. Initial identification of bacterial and archaeal species in each metagenomic sample was performed using this Kraken2 database. A custom Bowtie2 database, comprising only the core genes of identified species, was then established for refined analysis. Samples were mapped against this database, employing a compassionate option and a phred33 quality threshold. The mapping output was processed with SAMtools, and BEDtools were used to measure coverage. A stringent threshold of at least 25% coverage of core genes was set to confirm species identity using a specialized script. Species prevalence was determined based on the read count, normalized by the total core gene length per species.

4. Functional profiling based on shotgun metagenomics data

Functional profiling involved aligning each read to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database via DIAMOND. The initial DIAMOND database was built from a KEGG FASTA file containing ortholog amino acid sequences. BLASTx parameter in DIAMOND translated each metagenomic read into amino acid sequences across six reading frames for database comparison. The top-scoring KEGG hit was selected for each read. The quantification of KEGG orthologs was followed by MinPath analysis for predicting the presence of KEGG pathways.

5. Statistical analysis

The significance of differences in categorical variables between groups was determined using chi-square or Fisher’s exact tests. Continuous variables were analyzed by paired T-test or Wilcoxon’s signed-rank test, depending on whether the data had a normal distribution. p-values of less than 0.05 were considered to indicate statistical significance. The Kaplan-Meier method was used to evaluate progression-free survival (PFS), estimated from the diagnosis to the date of disease progression or all-cause death. The Kaplan-Meier method was used to construct survival curves, and the log-rank test was used to compare these curves between groups. Statistical analyses for clinical data were performed using SPSS ver. 25.0.

Microbiome analysis was performed using the Ez-Mx platform (CJ Bioscience Inc.). Alpha diversity (Shannon diversity) and beta diversity (Bray-Curtis dissimilarity) analyses were conducted using the ‘scikit-bio’ Python package. Bray-Curtis dissimilarity-based principal coordinate plot analysis was done on species profiles. Differential abundance in microbial species was analyzed using ANCOM-BC, focusing on taxa with over 10% prevalence. Functional biomarkers were examined through MaAsLin 2. Additionally, linear regression assessed the impact of clinical factors on the microbiome, and correlations between gut bacteria abundance and blood cytokine levels were analyzed using Spearman’s rank correlation in PRISM9 software.

6. Comparative analysis with ND-DLBCL

Whole genome shotgun metagenomic sequence data from ND-DLBCL patients were downloaded from a previous study (PRJNA928211). To compare the microbiomes of ENKTL patients, a dataset comprising 40 age- and sex-matched ND-DLBCL patients and 40 HCs was selected.

7. Data availability

Data from this study can be accessed in the National Center for Biotechnology Information repository and Sequence Read Archive database under the accession numbers PRJNA1043252.

Results

1. Clinical characteristics of ENKTL patients

The baseline clinical characteristics of the 40 patients at diagnosis are summarized in Table 1. More patients were male (n=24, 60.0%) than female (n=16, 40.0%). The median age was 50 (range, 21 to 82 years). Twenty-eight patients (70.0%) were younger than 60. Most patients experienced B-symptoms at the diagnosis (n=33, 82.5%). Twenty-five patients (62.5%) had detectable serum EBV DNA at diagnosis. More than half of the patients (n=32, 80.0%) presented with a limited stage (stage I or II), and the number of patients with high PINK or PINK-E measured at diagnosis was 17.5% (n=7) and 32.5% (n=13), respectively. Expression of PD-L1 ≥ 11 was identified in 23 cases (n=23/40, 59.0%). The number of patients who experienced early relapse within 12 months was 17 (42.5%).

2. Assessment of general microbiome between ENKTL and HCs

We compared the microbiome distribution at the phylum level between the ENKTL patients and the HCs. The intestinal microbiome predominantly comprised the phyla Firmicutes, Bacteroides, Actinobacteria, and Proteobacteria (Fig. 1A). Notably, there was a marked increase in the relative abundance of Proteobacteria in ENKTL patients compared to HCs, whereas the levels of other phyla did not exhibit significant changes (Fig. 1A). To delineate the distinct microbiome features of patients with ENKTL, we conducted a diversity analysis and compared it to the HC microbiome. The alpha diversity was particularly reduced in the ENKTL group compared to the HC (p=0.008) (Fig. 1B). Beta diversity in the ENKTL group significantly altered overall community composition compared to HC (p < 0.001) (Fig. 1C). The abundance of Enterobacteriaceae appeared to be a major force driving the variation in beta diversity, and there was a significant difference between the ENKTL and HC groups (p < 0.001) (Fig. 1D). A random forest model using family level taxonomic data established a commendable area under curve (0.87±0.03) (Fig. 1E), with Enterobacteriaceae as the feature with the greatest importance in the ENKTL group (Fig. 1F).

3. Functional alterations in the ENKTL microbiome

According to ANCOM-BC analysis, the Enterobacteriaceae family, especially the genera Escherichia, Citrobacter, and Enterobacter, showed higher abundance in the ENKTL group compared to the HC. Conversely, Akkermansia muciniphila, Paratractidigestivibacter faecalis, Coprococcus QULM, and Intestinimonas butyriciproducens, which are recognized as essential anaerobic gut symbionts for the biosynthesis of short-chain fatty acids (SCFAs), were found to be decreased in the ENKTL group compared to the HC (Fig. 2A). Specific characteristics in the ENKTL microbiome identified by Maaslin2 for microbial functional analysis included the upregulation of pathways linked to aerobic metabolism and aromatic compound degradation. Downregulated metabolic pathways in the ENKTL group were associated with the biosynthesis of butyrate, acetate, and folate (Fig. 2B).

Notably, the pathway exhibiting the greatest increase was ethanolamine utilization (PWY0-1477). In analysis at the gene level, we discovered that genes within the Eut operon, pivotal for ethanolamine metabolism, were significantly elevated in ENKTL (Fig. 2C, S1A Fig.). In contrast, the pathway undergoing the most substantial decrease was acetyl-CoA fermentation to butanoate (PWY-5676). Further analysis at the gene level showed a significant decline in genes essential for the biosynthesis of butyrate, including Thiolase (Thl; K00626), β-hydroxybutyryl-CoA dehydrogenase (Hdb; K00074), crotonase (Cro; K01252), butyryl-CoA dehydrogenase (Bcd; K00248), phosphate butyryltransferase (Ptb; K00634), butyrate kinase (Buk; K00929), and butyryl-CoA: acetate CoA transferase (But; K19709) (Fig. 2D, S1 Fig.).

4. Influence of microbial dysbiosis on clinical outcomes

We next explored the association between microbiome alterations observed in patients with ENKTL and clinical features. Redundancy analysis was conducted using 10 families with the highest importance identified in the random forest analysis, alongside clinical features (Fig. 3A). Interestingly, Enterobacteriaceae was found to be associated with not only patient status (serum CRP, stage, PINK, and PINK-E) but also clinical outcomes (early relapse, short PFS). Enterobacteriaceae was abundant in patients who experienced an early relapse within 12 months (Fig. 3B). In beta diversity analysis (Fig. 3C), the overall microbiome composition of patients who maintained a response was not significantly different from that of HCs (p=0.309), but there was a significant difference in patients who experienced early relapse (p < 0.001). Patients with a high abundance of Enterobacteriaceae showed significantly shorter PFS than those without (not reaching the median vs. 5.2 months; 95% confidence interval [CI], 4.2 to 6.2; p=0.001) (Fig. 3D).

Patients with an advanced-stage (stage III/IV) presented severe Enterobacteriaceae-related dysbiosis compared to limited-stage (stage I/II) patients in beta diversities (p < 0.001) (S2A and S2B Fig.). The advanced-stage patients showed poorer PFS than the limited-stage patients (not reached [NR] vs. 2.5 months; 95% CI, 1.3 to 3.7; p=0.006) (S2C Fig.). We categorized the patient group according to stages and abundance of Enterobacteriaceae to determine its influence on PFS. In patients with the same limited stage, those with low Enterobacteriaceae abundance showed better PFS than those with a higher abundance (NR vs. 11.2 months; 95% CI, 2.8 to 19.6). However, it was impossible to assess PFS in patients with an advanced stage according to Enterobacteriaceae abundance because no patients had a low abundance (2.5 months; 95% CI, 1.3 to 3.7; p < 0.001) (Fig. 3E). We used the chi-square test to explore the potential clinical indices that may be influenced by elevated levels of intestinal Enterobacteriaceae in ENKTL patients. The Enterobacteriaceae family demonstrated high PINK (p=0.003) and early relapse incidence (p=0.024) (Table 2). Based on linear regression analysis, clinical presentations such as elevated CRP, high PINK, and high PINK-E scores significantly correlated with increases in the relative abundance of Enterobacteriaceae (Fig. 3F). In univariate and multivariate analysis, the abundance of Enterobacteriaceae determined a significant association with PFS (p=0.047) (Table 3).

5. Similarities and differences in dysbiosis across lymphoma subtypes

Based on a previous study on ND-DLBCL microbiome data, this study analyzed the similarities and disparities in the intestinal microbial communities by comparing HCs, ND-DLBCL, and ND-ENKTL. The alpha diversity was markedly lower in ND-ENKTL and ND-DLBCL patients than in HCs (Fig. 4A). Moreover, ND-ENKTL and ND-DLBCL had compositions distinct from HCs in the beta diversity analysis (p < 0.001) (Fig. 4B). The microbial composition of ND-ENKTL presented a closer resemblance to DLBCL than to HCs. However, comparing ND-ENKTL and ND-DLBCL showed a notable statistical difference in the beta diversity (permutational multivariate analysis of variance, p=0.032) (S3A Fig.). While the relative abundance of Enterobacteriaceae at the family level was comparable between ENKTL and DLBCL patients (p=0.140) (Fig. 4C), differences emerged in the sub-composition within this detailed analysis (Fig. 4D). The ENKTL group demonstrated a relative abundance in Escherichia, whereas DLBCL was characterized by a predominance of Enterobacter and Citrobacter (S3B Fig). ANCOM-BC analysis was used to identify potential species-level biomarkers for ENKTL and DLBCL patients, Escherichia coli, Escherichia fergusonii, and Escherichia CP040443 were the most indicative biomarkers for ENKTL patients (S3C Fig.).

Consequently, we postulated that, among the Enterobacteriaceae family, Escherichia would significantly impact ENKTL (Fig. 4E). ENKTL patients with a high abundance of Escherichia exhibited an inferior PFS compared to those with a low abundance (NR vs. 5.2 months; 95% CI, 3.8 to 6.6; p=0.003) (Fig. 4F). Furthermore, the chi-square test revealed an association between elevated Escherichia levels with PINK (p=0.020) and PD-L1 (p=0.016) (Table 2). Linear regression analysis demonstrated a significant association between Escherichia abundance and PD-L1 levels (p=0.025) (Fig. 4G). However, no correlation with PD-L1 was observed for Enterobacteriaceae at the family level (p=0.571) (Fig. 4H).

Discussion

Mounting evidence indicates that significant microbiome and microbial functions facilitate the development of various cancers through direct or indirect immune modulation [14,15]. The association between the gut microbiome and lymphoid malignancies is explored less than in other cancer types [14,16]. Given that ENKTL stems from compromised innate immune surveillance and presents highly immunogenic characteristics [17,18], a substantial correlation is anticipated to exist between compromised intestinal barrier integrity, where innate immunity originates, and dysbiosis in the intestinal microbial community. According to our data, the ENKTL group presented higher Enterobacteriaceae at the family level and Escherichia, Citrobacter, and Enterobacter at the species level than the HC. These microbial compositions contributed to the decline of SCFA, producing strict anaerobic intestinal commensals in the ENKTL group. Additionally, the findings showed that Enterobacteriaceae-related dysbiosis affects the disease state and treatment outcome. Therefore, we would like to emphasize the need for the continued exploration of the ENKTL microbiome as a biomarker.

The relevance of gut microbiome dysbiosis to the clinical outcome and prognosis of cancer patients underscores the importance of understanding its relationship with ENKTL. A single gut microbiome study of 30 ND-ENKTL patients identified the relative abundance of the species Streptococcus parasanguinis and Romboutsia timonensis as biomarkers, and these microbiomes also have links to several diseases such as pancreatic cancer, Crohn disease, and liver disease [9]. Moreover, a previous study, which analyzed the nasal microbiome using 16s rRNA sequencing, suggested that the genus Corynebacterium and Staphylococcus represented a promising biomarker in patients with ENKTL [8]. Based on our shotgun sequencing of the gut microbiome, Enterobacteriaceae at the family level and Escherichia, Citrobacter, and Enterobacter at the species level were dominant in ND-ENKTL. Although these diverse microbiome findings may provide evidence that microbial dysbiosis plays an important role in the development of ENKTL, further studies are needed to elucidate the complex mechanisms of the bacterial-mediated development of lymphoma based on unified samples and analysis methods.

Alpha diversity analysis indicates the robustness of a microbial ecosystem, while beta diversity offers a measure to compare the overall similarity between different microbiome communities [19]. According to previous studies, low alpha diversity and differences in beta diversity are general gut microbiome features in cancer patients [20]. Research on various solid tumors has shown that patients exhibit lower alpha diversity and significantly different beta diversity [20]. Notably, alpha diversity is reported to be lower in non-responders than responders to anti–programmed death-1 immunotherapy [21]. In lymphoma studies, ND-DLBCL also demonstrated lower alpha diversity and significantly different beta diversity compared to HCs [22]. Additionally, a study on B-cell non-Hodgkin lymphoma showed that nonresponders had lower alpha diversity and significantly different beta diversity than responders [23]. In this study, we report that consistent with previous research, the gut microbiomes of ND-ENKTL patients exhibit lower alpha diversity and differing beta diversity compared to HCs. The primary driving force behind the differences in beta diversity was the Enterobacteriaceae family, which was cross-validated through analysis with a random forest model trained on family-level taxonomic data. In comparing cancer treatment responses in ND-ENKTL, there was no difference in alpha diversity between early relapse and late relapse groups, but there was a difference in beta diversity, with Enterobacteriaceae emerging as a notable biomarker.

According to community-level functional analysis, the gut microbiome of ENKTL patients displayed a significant increase in pathways related to aerobic substance metabolism and the degradation of aromatic compounds. Healthy human hosts control the oxygen concentration in the intestines, resulting in a nearly oxygen-free environment within the colon [24]. The presence of genes associated with high levels of aerobic metabolism in the microbiome of ENKTL patients suggests a failure to maintain an anaerobic environment in the gut. This failure may lead to an ecological advantage for facultative anaerobes such as Enterobacteriaceae that utilize aerobic respiration [25]. Additionally, the elevated presence of genes related to ethanolamine utilization indicates a higher presence of ethanolamine in the gut of ENKTL patients, which is commonly supplied in abundance following the destruction of intestinal epithelial cells [26]. Conversely, decreased functions include those related to the biosynthesis of SCFAs and folate. SCFAs contribute to lowering intestinal pH and oxygen concentration, thus preventing the colonization of pathogenic bacteria [27], whereas folate is essential for regulating the normal immune response of the host [28]. Decreased microbial metabolites present greater challenges in restoring microbial ecology and intestinal environment balance. Therefore, we suggest that ENKTL may indirectly or directly contribute, through yet unidentified pathways, to damaging the integrity of intestinal epithelial cells and the ability to regulate intestinal oxygen concentration, thereby inducing a bloom of Enterobacteriaceae and a decrease in SCFA and folate-producing bacteria. This gut microbial functional dysbiosis may contribute to a vicious cycle, which contributes to a continual microbial imbalance.

The abundance of intestinal Enterobacteriaceae in ENKTL patients is significantly linked with early relapse and dismal PFS. Nevertheless, given the observational nature of our study, elucidating the exact mechanism by which alterations in the microbiome influenced the treatment outcomes remains a challenge. Consequently, we undertook an indirect approach, exploring the relationship between Enterobacteriaceae abundance and clinical parameters that might affect patient prognosis. Through linear regression analysis, the factors that demonstrated a significant association with Enterobacteriaceae abundance included CRP, PINK, and PINK-E. CRP is a biomarker of acute inflammatory response and correlates with the proinflammatory cytokine interleukin-6 [29,30]. Elevated CRP levels are associated with advanced disease and an unfavorable prognosis among patients with various cancers [31,32]. We could not directly analyze the gut microbiome and systemic immune environment through cytokine analysis. Nevertheless, although the results do not clarify the precedence between disease status and microbiome dysbiosis, they emphasize the role of Enterobacteriaceae in fostering an environment conducive to heightened disease burden and elevated CRP levels.

Previously, our group reported that Proteobacteria at the phylum and Enterobacteriaceae at the family level were found in the gut microbiome of ND-DLBCL, and Enterobacteriaceae abundance showed a close correlation with poor treatment and survival outcomes [10]. The microbiome patterns observed in ENKTL and DLBCL were similar to Enterobacteriaceae-related dysbiosis at the family taxonomic level, with no discernible variance in relative abundance. Despite the constrained pathological resemblance between ENKTL and DLBCL, the recurring occurrence of elevated Enterobacteriaceae levels indicates that similar trends may be identifiable within other lymphoma subtypes. This proposition calls for additional investigation, prompting the need for microbiome assessments across a wider spectrum of lymphoma subtypes. In the foreseeable future, we are procuring samples to conduct microbial community analysis on patients diagnosed with follicular lymphoma and peripheral T-cell lymphoma not otherwise specified. Subsequently, we intend to examine the microbial community characteristics in lymphoma patients by leveraging the insights derived from microbiome analysis across diverse lymphoma subtypes.

Despite the microbiomes of patients with ENKTL or DLBCL showing considerable similarities, their compositions were not identical. At the family level, there was no difference in the abundance of Enterobacteriaceae, but there were significant differences in the sub-taxonomic level composition within Enterobacteriaceae. The microbiome of ENKTL was predominantly associated with Escherichia, while Citrobacter and Enterobacter dominated DLBCL. Interestingly, clinical outcomes associated with Enterobacteriaceae in ENKTL patients were found to be related only to Escherichia. Escherichia demonstrated a significant relationship with PFS in ENKTL patients, with neither Enterobacter nor Citrobacter exhibiting such an association. Furthermore, Escherichia exhibited a pronounced correlation with PD-L1 expression level. According to a previous study, ENKTL gene profiling data analysis has identified heightened activation of the JAK/STAT pathway within ENKTL cells, resulting in augmented signal transducer and activator of transcription 3 activation of PD-L1 and, consequently, an elevation in PD-L1 expression [33,34]. It seems dysbiosis in the ENKTL gut microbiome, which originated with E. coli, correlates with elevated PD-L1 expression and can escape immune surveillance.

In conclusion, we analyzed the gut microbiome in ND-ENKTL to identify specific microbial compositions and explore their potential as biomarkers. Our data shows that ND-ENKTL exhibits gut microbial dysbiosis characterized by low alpha diversity and increased Enterobacteriaceae. The functional analysis revealed an increase in genes related to aerobic respiration and ethanolamine utilization and a decrease in genes associated with the biosynthesis of butyrate and folate. However, further research is needed on the biological effects of metabolites in ND-ENKTL. Although the microbiomes of ND-ENKTL and ND-DLBCL are similar in terms of increased Enterobacteriaceae, ND-ENKTL shows a predominance of Escherichia, which was significantly associated with PD-L1 expression and shorter PFS. This study provides an opportunity to understand the microbial composition in ENKTL and its potential as a biomarker, but more research is needed to understand the interaction between lymphoma and the microbiome.

Electronic Supplementary Material

Supplementary materials are available at Cancer Research and Treatment website (https://www.e-crt.org).

crt-2024-675_S1_Fig.pdf

crt-2024-675_S2_Fig.pdf

crt-2024-675_S3_Fig.pdf

NOTES

Ethical Statement

This study was approved by the Institutional Review Board of Samsung Medical Center (approval number: 2016-11-040). It was conducted in accordance with the ethical principles of the Declaration of Helsinki and the Korea Good Clinical Practice guidelines. All participants provided written informed consent before taking part in the study.

Author Contributions

Conceived and designed the analysis: Kim SJ, Kim WS.

Collected the data: Yoon SE, Kang W, Cho J, Chalita M, Lee JH, Hyun DW, Kim H, Kim SJ, Kim WS.

Contributed data or analysis tools: Yoon SE, Kang W, Cho J, Chalita M, Lee JH, Hyun DW, Kim H, Kim SJ, Kim WS.

Performed the analysis: Yoon SE, Kang W.

Wrote the paper: Yoon SE, Kang W.

Conflicts of Interest

Conflict of interest relevant to this article was not reported.

Funding

This research was supported by a grant from the Korean Health Technology R&D Project through the Korean Health Industry Development Institute, funded by the Ministry of Health & Welfare, Republic of Korea (HR20C0025), a National Research Foundation of Korea grant funded by the Korean government (2022R1F1A1064058), and the Bio&Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2023-00222838).

Fig. 1.

(A) Predominant microbial phyla in healthy control (HC) and extranodal natural killer/T-cell lymphoma (ENKTL), including Firmicutes, Bacteroidetes, Actinobacteria, and Proteobacteria. Phyla with less than 1% abundance are grouped as ‘Others’. (B) Alpha diversity assessed using the Shannon index in HC and ENKTL cohorts. (C) Beta diversity through Bray-Curtis dissimilarity between the groups. (D) Enterobacteriaceae abundance comparison in HC and ENKTL. (E) ROC curves analysis from a random forest model for patient stratification criteria, depicting true positive rate against false positive rate. (F) Identification of top 30 taxa as key features in the random forest model relevant to HC and ENKTL differentiation. AUC, area under curve; SD, standard deviation. ns, not significant; ^**p < 0.01, ^***p < 0.001.

Fig. 2.

(A) Species-level gut microbial taxonomic biomarkers for extranodal natural killer/T-cell lymphoma (ENKTL) and healthy control (HC). The top 10 biomarkers for each group are displayed, with ENKTL markers in red and HC markers in blue. Biomarkers identified by using ANCOM-BC analysis. (B) Volcano plot illustrating pathway-level functional biomarkers. Positive x-axis values represent biomarkers for ENKTL (marked in red), whereas negative values indicate HC biomarkers (marked in blue). Statistically significant markers are highlighted. Functional biomarkers analyzed using MaAsLin 2. (C, D) Statistical analysis of the cumulative reads per kilobase million (RPKM) of genes constituting major pathways representative of ENKTL and HC. (C) Ethanolamine utilization pathway. (D) Butyrate biosynthesis pathway. ^**p < 0.01, ^***p < 0.001.

Fig. 3.

(A) Redundancy analysis (RDA) on the top 10 families with high feature importance from random forest analysis, in correlation with key clinical features of extranodal natural killer/T-cell lymphoma (ENKTL). Green dots represent individual ENKTL patients, blue dots denote families, and red arrows indicate clinical features. (B) Comparative analysis of Enterobacteriaceae abundance among healthy control (HC), maintained response, and early relapse groups. (C) Beta diversity assessment (Bray-Curtis) for HC, maintained response, and early relapse groups. (D, E) Progression-free survival (PFS) comparison in ENKTL patients divided into top 50% and bottom 50% based on Enterobacteriaceae abundance. The upper panel displays overall PFS (D); the lower panel shows PFS further stratified into stage I/II and III/IV within each subgroup (E). (F) Linear regression plots demonstrating the relationship between Enterobacteriaceae abundance and clinical biomarkers C-reactive protein (CRP), prognosis index of natural killer cell lymphoma (PINK), and PINK-E. BMI, body mass index; EBV, Epstein-Barr virus; LDH, lactate dehydrogenase; OS, overall survival; PFS, progression-free survival. ns, not significant; ^**p < 0.01, ^***p < 0.001.

Fig. 4.

(A) Comparative analysis of alpha diversity (Shannon index) among healthy control (HC), diffuse large B-cell lymphoma (DLBCL), and extranodal natural killer/T-cell lymphoma (ENKTL) groups. (B) Beta diversity comparison (Bray-Curtis) across HC, DLBCL, and ENKTL groups. (C) Relative abundance of Enterobacteriaceae in DLBCL and ENKTL groups. (D) Comparison of the genus composition within the Enterobacteriaceae family between DLBCL and ENKTL groups. (E) Progression-free survival (PFS) analysis in ENKTL patients categorized into top 50% and bottom 50% based on Escherichia abundance. (F) PFS comparison in ENKTL patients stratified by relative abundance of Escherichia into top 50% and bottom 50%, further subdivided into stage I/II and III/IV. (G, H) Linear regression analysis illustrating the relationship between Escherichia relative abundance and serum programmed cell death-ligand-1 (PD-L1) levels (G), and between Enterobacteriaceae relative abundance and serum PD-L1 levels (H).

Table 1.

Baseline characteristics

Variable	No. (%) (n=40)
Sex
Male	24 (60.0)
Female	16 (40.0)
Age (yr)
≤ 60	28 (70.0)
> 60	12 (30.0)
ECOG PS
0-1	33 (82.5)
≥ 2	7 (17.5)
B-symptom
Presence	38 (95.0)
Absence	2 (5.0)
LDH
Normal	24 (60.0)
Elevation	16 (40.0)
CRP (n=36)
Normal	30 (83.3)
Elevation	6 (16.7)
Circulating EBV detection
Absence	15 (37.5)
Elevation	25 (62.5)
Stage
I/II	32 (80.0)
III/IV	8 (20.0)
PINK
0-1	33 (82.5)
≥ 2	7 (17.5)
PINK-E
0-1	27 (67.5)
≥ 2	13 (32.5)
PD-L1^a) (n=39)
≤ 10	16 (41.0)
> 10	23 (59.0)
TIME^b) (n=39)
IT+IE-A	25 (64.1)
IE-B+IS	14 (35.9)
Early relapse^c)
No	23 (57.5)
Yes	17 (42.5)

CRP, C-reactive protein; EBV, Epstein-Barr virus; ECOG PS, Eastern Cooperative Oncology Group performance status; ENKTL, extranodal natural killer/T-cell lymphoma; IE-A, immune evasion-A; IE-B, immune evasion-B; IHC, immunohistochemistry; IS, immune silence; IT, immune tolerance; LDH, lactate dehydrogenase; PD-L1, programmed death-ligand 1; PINK, prognostic index of natural killer cell lymphoma; PINK-E, prognostic index of natural killer cell lymphoma with EBV; TIME, tumor immune microenvironment.

^a) Programmed cell death-ligand-1 (PD-L1) expression data using a PD-L1 IHC 22C3 pharmDx kit and the Dako ASL48 platform. PD-L1 positivity was defined as membranous PD-L1 expression by more than 10% of tumor cells, regardless of staining intensity,

^b) TIME of ENKTL based on the IHC markers FoxP3, PD-L1, and CD68. We classified four distinct TIME subtypes: IT, IE-A, IE-B, and IS,

^c) Early relapse was defined as disease progression within 12 months.

Table 2.

The comparison of baseline variables according to the microbiome at the family and genus levels

	Escherichia (genus)			Enterobacteriaceae (family)
	Low (n=32)	High (n=8)	p-value	Low (n=21)	High (n=19)	p-value
Sex
Male	18 (75.0)	6 (25.0)	0.439	12 (50.0)	12 (50.0)	0.755
Female	14 (87.5)	2 (12.5)		9 (56.3)	7 (43.8)
Age (yr)
≤ 60	6 (21.4)	22 (78.6)	> 0.99	14 (50.0)	14 (50.0)	0.736
> 60	2 (16.7)	10 (83.3)		5 (41.7)	7 (58.3)
ECOG PS
0-1	26 (78.8)	7 (21.2)	> 0.99	17 (51.5)	16 (48.5)	> 0.99
≥ 2	6 (85.7)	1 (14.3)		4 (57.1)	3 (42.9)
B-symptom
Presence	30 (78.9)	8 (21.1)	> 0.99	19 (50.0)	19 (50.0)	0.488
Absence	2 (100)	0		2 (100)	0
LDH
Normal	19 (79.5)	5 (20.8)	> 0.99	15 (62.5)	9 (37.5)	0.196
Elevation	13 (81.3)	3 (18.8)		6 (37.5)	10 (62.5)
CRP
Normal	24 (80.0)	6 (20.0)	0.596	17 (56.7)	13 (43.3)	0.391
Elevation	4 (66.7)	2 (33.3)		2 (33.3)	4 (66.7)
Circulating EBV detection
Absence	12 (80.0)	3 (20.0)	> 0.99	10 (66.7)	5 (33.3)	0.204
Elevation	20 (80.0)	5 (20.0)		11 (44.0)	14 (56.0)
Stage
I/II	28 (87.5)	4 (12.5)	0.037	21 (65.6)	11 (34.4)	0.001
II/IV	4 (50.0)	5 (50.0)		0	8 (100)
PINK
0-1	29 (87.9)	4 (12.1)	0.020	21 (63.6)	12 (36.4)	0.003
≥ 2	3 (42.9)	4 (57.1)		0	7 (100)
PINK-E
0-1	23 (85.2)	4 (14.8)	0.400	17 (63.0)	10 (37.0)	0.091
≥ 2	9 (69.2)	4 (30.8)		4 (30.8)	9 (69.2)
PD-L1^a) (n=39)
≤ 10	0	16 (100)	0.016	5 (31.3)	11 (68.8)	0.192
>10	7 (30.4)	16 (69.6)		13 (56.5)	10 (43.5)
TIME^b) (n=39)
IT+IE-A	19 (76.0)	6 (24.0)	0.237	12 (48.0)	13 (52.0)	0.504
IE-B+IS	13 (92.9)	1 (7.1)		9 (64.3)	5 (35.7)
Early relapse^c)
No	20 (87.0)	3 (13.0)	0.250	16 (69.6)	7 (30.4)	0.024
Yes	12 (70.6)	5 (29.4)		5 (29.4)	12 (70.6)

Values are presented as number (%). CRP, C-reactive protein; EBV, Epstein-Barr virus; ECOG PS, Eastern Cooperative Oncology Group performance status; ENKTL, extranodal natural killer/T-cell lymphoma; IE-A, immune evasion-A; IE-B, immune evasion-B; IHC, immunohistochemistry; IS, immune silence; IT, immune tolerance; LDH, lactate dehydrogenase; PD-L1, programmed death-ligand 1; PINK, prognostic index of natural killer cell lymphoma; PINK-E, prognostic index of natural killer cell lymphoma with EBV; TIME, tumor immune microenvironment.

^b) TIME of ENKTL based on the IHC markers FoxP3, PD-L1, and CD68. We classified four distinct TIME subtypes: IT, IE-A, IE-B, and IS,

^c) Early relapse was defined as disease progression within 12 months.

Table 3.

Univariate and multivariate analysis of PFS

	Univariate			Multivariate
	HR	95% CI	p-value	HR	95% CI	p-value
Sex
Male vs. female	0.672	0.285-1.587	0.365	-	-	-
Age (yr)
≤ 60 vs. > 60	1.247	0.512-3.035	0.627	-	-	-
ECOG-PS
0-1 vs. ≥ 2	0.362	0.085-1.549	0.171	-	-	-
B-Symptom
Presence vs. absence	0.044	0.000-43.966	0.375	-	-	-
LDH
Normal vs. elevation	1.662	0.726-3.803	0.229	-	-	-
CRP
Normal vs. elevation	2.147	0.706-6.523	0.178	-	-	-
Stage
I/II vs. III/IV	11.513	3.961-33.461	< 0.001	1.939	0.141-26.600	0.620
PINK
0-1 vs. ≥ 2	9.833	3.442-28.379	< 0.001	0.991	0.115-8.525	0.993
PINK-E
0-1 vs. ≥ 2	5.217	2.235-12.177	< 0.001	3.777	0.921-15.495	0.065
PD-L1
≤ 10 vs. > 10	1.077	0.460-2.523	0.864	-	-	-
Enterobacteriaceae (family)
Low vs. high presentation	0.259	0.108-0.621	0.002	0.317	0.102-0.986	0.047
Escherichia (genus)
Low vs. high presentation	0.227	0.091-0.564	0.001	0.713	0.208-2.441	0.590

CI, confidence interval; CRP, C-reactive protein; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; LDH, lactate dehydrogenase; PD-L1, programmed death-ligand 1; PFS, progression-free survival; PINK, prognostic index of natural killer cell lymphoma; PINK-E, prognostic index of natural killer cell lymphoma with Epstein-Barr virus.

REFERENCES

1. Damania B, Kenney SC, Raab-Traub N. Epstein-Barr virus: biology and clinical disease. Cell. 2022;185:3652–70. Article PubMed PMC
2. Tse E, Fox CP, Glover A, Yoon SE, Kim WS, Kwong YL. Extranodal natural killer/T-cell lymphoma: an overview on pathology and clinical management. Semin Hematol. 2022;59:198–209. Article PubMed
3. Lin GW, Xu C, Chen K, Huang HQ, Chen J, Song B, et al. Genetic risk of extranodal natural killer T-cell lymphoma: a genome-wide association study in multiple populations. Lancet Oncol. 2020;21:306–16. PubMed
4. Li Z, Xia Y, Feng LN, Chen JR, Li HM, Cui J, et al. Genetic risk of extranodal natural killer T-cell lymphoma: a genome-wide association study. Lancet Oncol. 2016;17:1240–7. PubMed PMC
5. Vose J, Armitage J, Weisenburger D; International T-Cell Lymphoma Project. International peripheral T-cell and natural killer/T-cell lymphoma study: pathology findings and clinical outcomes. J Clin Oncol. 2008;26:4124–30. Article PubMed
6. Yoon SE, Song Y, Kim SJ, Yoon DH, Chen TY, Koh Y, et al. Comprehensive analysis of peripheral T-cell and natural killer/T-cell lymphoma in Asian patients: a multinational, multicenter, prospective registry study in Asia. Lancet Reg Health West Pac. 2021;10:100126.Article PubMed PMC
7. Gupta VK, Paul S, Dutta C. Geography, ethnicity or subsistence-specific variations in human microbiome composition and diversity. Front Microbiol. 2017;8:1162.Article PubMed PMC
8. Shi Z, Li X, Wang X, Zhang L, Li L, Fu X, et al. Characteristics and clinical implications of the nasal microbiota in extranodal NK/T-cell lymphoma, nasal type. Front Cell Infect Microbiol. 2021;11:686595.Article PubMed PMC
9. Shi Z, Hu G, Li MW, Zhang L, Li X, Li L, et al. Gut microbiota as non-invasive diagnostic and prognostic biomarkers for natural killer/T-cell lymphoma. Gut. 2023;72:1999–2002. Article PubMed
10. Yoon SE, Kang W, Choi S, Park Y, Chalita M, Kim H, et al. The influence of microbial dysbiosis on immunochemotherapy-related efficacy and safety in diffuse large B-cell lymphoma. Blood. 2023;141:2224–38. Article PubMed PDF
11. Kim SJ, Yoon DH, Jaccard A, Chng WJ, Lim ST, Hong H, et al. A prognostic index for natural killer cell lymphoma after non-anthracycline-based treatment: a multicentre, retrospective analysis. Lancet Oncol. 2016;17:389–400. Article PubMed
12. Cho J, Kim SJ, Park WY, Kim J, Woo J, Kim G, et al. Immune subtyping of extranodal NK/T-cell lymphoma: a new biomarker and an immune shift during disease progression. Mod Pathol. 2020;33:603–15. Article PubMed PDF
13. Cheson BD, Pfistner B, Juweid ME, Gascoyne RD, Specht L, Horning SJ, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25:579–86. Article PubMed
14. Sepich-Poore GD, Zitvogel L, Straussman R, Hasty J, Wargo JA, Knight R. The microbiome and human cancer. Science. 2021;371:eabc4552Article PubMed PMC
15. Park EM, Chelvanambi M, Bhutiani N, Kroemer G, Zitvogel L, Wargo JA. Targeting the gut and tumor microbiota in cancer. Nat Med. 2022;28:690–703. Article PubMed PDF
16. Upadhyay Banskota S, Skupa SA, El-Gamal D, D’Angelo CR. Defining the role of the gut microbiome in the pathogenesis and treatment of lymphoid malignancies. Int J Mol Sci. 2023;24:2309.Article PubMed PMC
17. Hamerman JA, Ogasawara K, Lanier LL. NK cells in innate immunity. Curr Opin Immunol. 2005;17:29–35. Article PubMed
18. Bajor-Dattilo EB, Pittaluga S, Jaffe ES. Pathobiology of T-cell and NK-cell lymphomas. Best Pract Res Clin Haematol. 2013;26:75–87. Article PubMed PMC
19. Kers JG, Saccenti E. The power of microbiome studies: some considerations on which alpha and beta metrics to use and how to report results. Front Microbiol. 2021;12:796025.Article PubMed
20. Helmink BA, Khan MA, Hermann A, Gopalakrishnan V, Wargo JA. The microbiome, cancer, and cancer therapy. Nat Med. 2019;25:377–88. Article PubMed PDF
21. Gopalakrishnan V, Spencer CN, Nezi L, Reuben A, Andrews MC, Karpinets TV, et al. Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients. Science. 2018;359:97–103. PubMed
22. Yuan L, Wang W, Zhang W, Zhang Y, Wei C, Li J, et al. Gut microbiota in untreated diffuse large B cell lymphoma patients. Front Microbiol. 2021;12:646361.Article PubMed PMC
23. Diefenbach CS, Peters BA, Li H, Raphael B, Moskovits T, Hymes K, et al. Microbial dysbiosis is associated with aggressive histology and adverse clinical outcome in B-cell nonHodgkin lymphoma. Blood Adv. 2021;5:1194–8. Article PubMed PMC PDF
24. Rivera-Chavez F, Lopez CA, Baumler AJ. Oxygen as a driver of gut dysbiosis. Free Radic Biol Med. 2017;105:93–101. Article PubMed
25. Litvak Y, Byndloss MX, Baumler AJ. Colonocyte metabolism shapes the gut microbiota. Science. 2018;362:eaat9076Article PubMed PMC
26. Zeng MY, Inohara N, Nunez G. Mechanisms of inflammation-driven bacterial dysbiosis in the gut. Mucosal Immunol. 2017;10:18–26. Article PubMed PDF
27. Liu XF, Shao JH, Liao YT, Wang LN, Jia Y, Dong PJ, et al. Reglation of short-chain fatty acids in the immune system. Front Immunol. 2023;14:1186892.PubMed PMC
28. Mikkelsen K, Apostolopoulos V. Vitamin B12, folic acid, and the immune system. In: Mahmoudi M, Razaei N, editors. Nutrition and immunity. Springer; 2019. p. 103–14.
29. Li YJ, Li ZM, Xia Y, Huang JJ, Huang HQ, Xia ZJ, et al. Serum C-reactive protein (CRP) as a simple and independent prognostic factor in extranodal natural killer/T-cell lymphoma, nasal type. PLoS One. 2013;8:e64158Article PubMed PMC
30. Bao C, Zhou D, Zhu L, Qian W, Ye X. Increased serum level of interleukin-6 correlates with negative prognostic factors in extranodal NK/T-cell lymphoma. Transl Cancer Res. 2020;9:2378–89. Article PubMed PMC
31. Mitsunaga S, Ikeda M, Shimizu S, Ohno I, Takahashi H, Okuyama H, et al. C-reactive protein level is an indicator of the aggressiveness of advanced pancreatic cancer. Pancreas. 2016;45:110–6. Article PubMed
32. Troppan KT, Schlick K, Deutsch A, Melchardt T, Egle A, Stojakovic T, et al. C-reactive protein level is a prognostic indicator for survival and improves the predictive ability of the R-IPI score in diffuse large B-cell lymphoma patients. Br J Cancer. 2014;111:55–60. Article PubMed PMC PDF
33. Song TL, Nairismagi ML, Laurensia Y, Lim JQ, Tan J, Li ZM, et al. Oncogenic activation of the STAT3 pathway drives PD-L1 expression in natural killer/T-cell lymphoma. Blood. 2018;132:1146–58. Article PubMed PMC PDF
34. Li X, Cheng Y, Zhang M, Yan J, Li L, Fu X, et al. Activity of pembrolizumab in relapsed/refractory NK/T-cell lymphoma. J Hematol Oncol. 2018;11:15.Article PubMed PMC PDF

Figure & Data

REFERENCES

Citations

Citations to this article as recorded by

PubReader
ePub Link

Cite

CITE: export Copy Download Format; Close

Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:

RIS — For EndNote, ProCite, RefWorks, and most other reference management software
BibTeX — For JabRef, BibDesk, and other BibTeX-specific software

Include:

Citation for the content below
Citation and abstract for the content below

Clinical Impact of Microbiome Characteristics in Treatment-Naïve Extranodal NK/T-Cell Lymphoma Patients

Cancer Res Treat. 2025;57(2):597-611. Published online August 16, 2024

DOI: https://doi.org/10.4143/crt.2024.675

XML Download

Abstract
Introduction
Materials and Methods
Results
Discussion
Electronic Supplementary Material
NOTES
REFERENCES

Clinical Impact of Microbiome Characteristics in Treatment-Naïve Extranodal NK/T-Cell Lymphoma Patients

Fig. 1. (A) Predominant microbial phyla in healthy control (HC) and extranodal natural killer/T-cell lymphoma (ENKTL), including Firmicutes, Bacteroidetes, Actinobacteria, and Proteobacteria. Phyla with less than 1% abundance are grouped as ‘Others’. (B) Alpha diversity assessed using the Shannon index in HC and ENKTL cohorts. (C) Beta diversity through Bray-Curtis dissimilarity between the groups. (D) Enterobacteriaceae abundance comparison in HC and ENKTL. (E) ROC curves analysis from a random forest model for patient stratification criteria, depicting true positive rate against false positive rate. (F) Identification of top 30 taxa as key features in the random forest model relevant to HC and ENKTL differentiation. AUC, area under curve; SD, standard deviation. ns, not significant; **p < 0.01, ***p < 0.001.

Fig. 2. (A) Species-level gut microbial taxonomic biomarkers for extranodal natural killer/T-cell lymphoma (ENKTL) and healthy control (HC). The top 10 biomarkers for each group are displayed, with ENKTL markers in red and HC markers in blue. Biomarkers identified by using ANCOM-BC analysis. (B) Volcano plot illustrating pathway-level functional biomarkers. Positive x-axis values represent biomarkers for ENKTL (marked in red), whereas negative values indicate HC biomarkers (marked in blue). Statistically significant markers are highlighted. Functional biomarkers analyzed using MaAsLin 2. (C, D) Statistical analysis of the cumulative reads per kilobase million (RPKM) of genes constituting major pathways representative of ENKTL and HC. (C) Ethanolamine utilization pathway. (D) Butyrate biosynthesis pathway. **p < 0.01, ***p < 0.001.

Fig. 3. (A) Redundancy analysis (RDA) on the top 10 families with high feature importance from random forest analysis, in correlation with key clinical features of extranodal natural killer/T-cell lymphoma (ENKTL). Green dots represent individual ENKTL patients, blue dots denote families, and red arrows indicate clinical features. (B) Comparative analysis of Enterobacteriaceae abundance among healthy control (HC), maintained response, and early relapse groups. (C) Beta diversity assessment (Bray-Curtis) for HC, maintained response, and early relapse groups. (D, E) Progression-free survival (PFS) comparison in ENKTL patients divided into top 50% and bottom 50% based on Enterobacteriaceae abundance. The upper panel displays overall PFS (D); the lower panel shows PFS further stratified into stage I/II and III/IV within each subgroup (E). (F) Linear regression plots demonstrating the relationship between Enterobacteriaceae abundance and clinical biomarkers C-reactive protein (CRP), prognosis index of natural killer cell lymphoma (PINK), and PINK-E. BMI, body mass index; EBV, Epstein-Barr virus; LDH, lactate dehydrogenase; OS, overall survival; PFS, progression-free survival. ns, not significant; **p < 0.01, ***p < 0.001.

Fig. 4. (A) Comparative analysis of alpha diversity (Shannon index) among healthy control (HC), diffuse large B-cell lymphoma (DLBCL), and extranodal natural killer/T-cell lymphoma (ENKTL) groups. (B) Beta diversity comparison (Bray-Curtis) across HC, DLBCL, and ENKTL groups. (C) Relative abundance of Enterobacteriaceae in DLBCL and ENKTL groups. (D) Comparison of the genus composition within the Enterobacteriaceae family between DLBCL and ENKTL groups. (E) Progression-free survival (PFS) analysis in ENKTL patients categorized into top 50% and bottom 50% based on Escherichia abundance. (F) PFS comparison in ENKTL patients stratified by relative abundance of Escherichia into top 50% and bottom 50%, further subdivided into stage I/II and III/IV. (G, H) Linear regression analysis illustrating the relationship between Escherichia relative abundance and serum programmed cell death-ligand-1 (PD-L1) levels (G), and between Enterobacteriaceae relative abundance and serum PD-L1 levels (H).

Fig. 1.

Fig. 2.

Fig. 3.

Fig. 4.

Clinical Impact of Microbiome Characteristics in Treatment-Naïve Extranodal NK/T-Cell Lymphoma Patients

Variable	No. (%) (n=40)
Sex
Male	24 (60.0)
Female	16 (40.0)
Age (yr)
≤ 60	28 (70.0)
> 60	12 (30.0)
ECOG PS
0-1	33 (82.5)
≥ 2	7 (17.5)
B-symptom
Presence	38 (95.0)
Absence	2 (5.0)
LDH
Normal	24 (60.0)
Elevation	16 (40.0)
CRP (n=36)
Normal	30 (83.3)
Elevation	6 (16.7)
Circulating EBV detection
Absence	15 (37.5)
Elevation	25 (62.5)
Stage
I/II	32 (80.0)
III/IV	8 (20.0)
PINK
0-1	33 (82.5)
≥ 2	7 (17.5)
PINK-E
0-1	27 (67.5)
≥ 2	13 (32.5)
PD-L1^a) (n=39)
≤ 10	16 (41.0)
> 10	23 (59.0)
TIME^b) (n=39)
IT+IE-A	25 (64.1)
IE-B+IS	14 (35.9)
Early relapse^c)
No	23 (57.5)
Yes	17 (42.5)

	Escherichia (genus)			Enterobacteriaceae (family)
	Low (n=32)	High (n=8)	p-value	Low (n=21)	High (n=19)	p-value
Sex
Male	18 (75.0)	6 (25.0)	0.439	12 (50.0)	12 (50.0)	0.755
Female	14 (87.5)	2 (12.5)		9 (56.3)	7 (43.8)
Age (yr)
≤ 60	6 (21.4)	22 (78.6)	> 0.99	14 (50.0)	14 (50.0)	0.736
> 60	2 (16.7)	10 (83.3)		5 (41.7)	7 (58.3)
ECOG PS
0-1	26 (78.8)	7 (21.2)	> 0.99	17 (51.5)	16 (48.5)	> 0.99
≥ 2	6 (85.7)	1 (14.3)		4 (57.1)	3 (42.9)
B-symptom
Presence	30 (78.9)	8 (21.1)	> 0.99	19 (50.0)	19 (50.0)	0.488
Absence	2 (100)	0		2 (100)	0
LDH
Normal	19 (79.5)	5 (20.8)	> 0.99	15 (62.5)	9 (37.5)	0.196
Elevation	13 (81.3)	3 (18.8)		6 (37.5)	10 (62.5)
CRP
Normal	24 (80.0)	6 (20.0)	0.596	17 (56.7)	13 (43.3)	0.391
Elevation	4 (66.7)	2 (33.3)		2 (33.3)	4 (66.7)
Circulating EBV detection
Absence	12 (80.0)	3 (20.0)	> 0.99	10 (66.7)	5 (33.3)	0.204
Elevation	20 (80.0)	5 (20.0)		11 (44.0)	14 (56.0)
Stage
I/II	28 (87.5)	4 (12.5)	0.037	21 (65.6)	11 (34.4)	0.001
II/IV	4 (50.0)	5 (50.0)		0	8 (100)
PINK
0-1	29 (87.9)	4 (12.1)	0.020	21 (63.6)	12 (36.4)	0.003
≥ 2	3 (42.9)	4 (57.1)		0	7 (100)
PINK-E
0-1	23 (85.2)	4 (14.8)	0.400	17 (63.0)	10 (37.0)	0.091
≥ 2	9 (69.2)	4 (30.8)		4 (30.8)	9 (69.2)
PD-L1^a) (n=39)
≤ 10	0	16 (100)	0.016	5 (31.3)	11 (68.8)	0.192
>10	7 (30.4)	16 (69.6)		13 (56.5)	10 (43.5)
TIME^b) (n=39)
IT+IE-A	19 (76.0)	6 (24.0)	0.237	12 (48.0)	13 (52.0)	0.504
IE-B+IS	13 (92.9)	1 (7.1)		9 (64.3)	5 (35.7)
Early relapse^c)
No	20 (87.0)	3 (13.0)	0.250	16 (69.6)	7 (30.4)	0.024
Yes	12 (70.6)	5 (29.4)		5 (29.4)	12 (70.6)

	Univariate			Multivariate
	HR	95% CI	p-value	HR	95% CI	p-value
Sex
Male vs. female	0.672	0.285-1.587	0.365	-	-	-
Age (yr)
≤ 60 vs. > 60	1.247	0.512-3.035	0.627	-	-	-
ECOG-PS
0-1 vs. ≥ 2	0.362	0.085-1.549	0.171	-	-	-
B-Symptom
Presence vs. absence	0.044	0.000-43.966	0.375	-	-	-
LDH
Normal vs. elevation	1.662	0.726-3.803	0.229	-	-	-
CRP
Normal vs. elevation	2.147	0.706-6.523	0.178	-	-	-
Stage
I/II vs. III/IV	11.513	3.961-33.461	< 0.001	1.939	0.141-26.600	0.620
PINK
0-1 vs. ≥ 2	9.833	3.442-28.379	< 0.001	0.991	0.115-8.525	0.993
PINK-E
0-1 vs. ≥ 2	5.217	2.235-12.177	< 0.001	3.777	0.921-15.495	0.065
PD-L1
≤ 10 vs. > 10	1.077	0.460-2.523	0.864	-	-	-
Enterobacteriaceae (family)
Low vs. high presentation	0.259	0.108-0.621	0.002	0.317	0.102-0.986	0.047
Escherichia (genus)
Low vs. high presentation	0.227	0.091-0.564	0.001	0.713	0.208-2.441	0.590

Table 1. Baseline characteristics

Programmed cell death-ligand-1 (PD-L1) expression data using a PD-L1 IHC 22C3 pharmDx kit and the Dako ASL48 platform. PD-L1 positivity was defined as membranous PD-L1 expression by more than 10% of tumor cells, regardless of staining intensity,

TIME of ENKTL based on the IHC markers FoxP3, PD-L1, and CD68. We classified four distinct TIME subtypes: IT, IE-A, IE-B, and IS,

Early relapse was defined as disease progression within 12 months.

Table 2. The comparison of baseline variables according to the microbiome at the family and genus levels

TIME of ENKTL based on the IHC markers FoxP3, PD-L1, and CD68. We classified four distinct TIME subtypes: IT, IE-A, IE-B, and IS,