Skip to main content
Cancer Science logoLink to Cancer Science
. 2026 Nov 9:10.1111/cas.70570. Online ahead of print. doi: 10.1111/cas.70570

3‐Indoxyl Sulfate and O‐Acetylcarnitine Are Potential Novel Biomarkers for Early Diagnosis of Esophageal Cancer

Kenji Hamada 1,2, Yohei Ozawa 1,✉, Hiroshi Okamoto 1, Shinji Fukuda 3,4,5,6, Tomoyoshi Soga 3,7, Takaaki Abe 8, Hozumi Motohashi 2, Shohei Murakami 2,✉, Takashi Kamei 1
PMCID: PMC13748909  PMID: 43004740

ABSTRACT

Esophageal cancer is the seventh most common cancer worldwide, and early detection is crucial for improving long‐term survival. However, reliable noninvasive biomarkers for early‐stage disease remain limited. In this study, we aimed to identify novel biomarkers for early‐stage esophageal cancer through comprehensive metabolomic analyses of plasma and urine samples from patients and healthy individuals. Metabolomic profiling revealed that 3‐indoxyl sulfate (3‐IS) was consistently elevated in both plasma and urine of esophageal cancer patients. As esophageal cancer cells do not directly produce 3‐IS, its elevation may reflect cancer‐associated systemic metabolic alterations. O‐acetylcarnitine (O‐AC) was identified as the most significantly increased metabolite in plasma, and in vitro analyses suggested a possible association with tumor‐related metabolic changes. Receiver operating characteristic analyses (ROC) demonstrated that the combined plasma and urinary 3‐IS model achieved an area under the curve (AUC) of 0.863 in Stage I/II disease, whereas plasma O‐AC achieved an AUC of 0.902. Bootstrap resampling provided internal validation; however, the discovery and evaluation were performed within the same single‐center cohort. These results suggest that 3‐IS and O‐AC are promising candidate biomarkers that distinguish patients with esophageal cancer from healthy individuals, but further validation of their disease specificity and diagnostic performance is required in independent multicenter cohorts including clinically relevant disease‐control groups.

Keywords: 3‐Indoxyl sulfate, biomarkers, early diagnosis, esophageal cancer, O‐acetylcarnitine


Through comprehensive analysis of plasma and urine from patients with esophageal cancer and healthy controls, 3‐indoxyl sulfate and O‐acetylcarnitine were identified as metabolites significantly associated with esophageal cancer. These metabolites were elevated in patients and demonstrated promising diagnostic performance for early detection. NT, macroscopically non‐neoplastic tissue.

graphic file with name CAS-9999-0-g011.webp


Abbreviations

3‐IS

3‐indoxyl sulfate

AHR

aryl hydrocarbon receptor

AUC

area under the curve

BMI

body mass index

CEA

carcinoembryonic antigen

CE‐TOFMS

capillary electrophoresis time‐of‐flight mass spectrometry

ctDNA

circulating tumor DNA

ESI

electrospray ionization

FDR

false discovery rate

GLM

general linear model

LC–MS/MS

liquid chromatography–tandem mass spectrometry

MRM

multiple reaction monitoring

NT

macroscopically non‐neoplastic tissue

O‐AC

O‐acetylcarnitine

qRT‐PCR

quantitative real‐time polymerase chain reaction

ROC

receiver operating characteristic

SCC

squamous cell carcinoma

1. Introduction

Esophageal cancer is the seventh most common cancer worldwide, with 600,000 new cases reported in 2020 [1]. The global 5‐year survival rate is poor, ranging from 15% to 25%, and in Japan, the 5‐year survival rate is also poor at 40% [2, 3]. Esophageal cancer is a malignant tumor with high recurrence and metastasis. After radical resection, the total recurrence rate is 36%–56%, and the median time to recurrence ranges from 10 to 12 months [4, 5]. Particularly in advanced cancer (Stage III/IV), the recurrence rate is as high as 64.2% [6]. Esophageal cancer can be generally divided into two histological subtypes: squamous cell carcinoma (SCC) and adenocarcinoma. Smoking and alcohol consumption are well‐known risk factors for SCC [7], while obesity is a risk factor for adenocarcinoma [8, 9]. The low rates of survival and complete cure are highly caused by difficulty in early diagnosis, due to a lack of initial symptoms and no well‐established detection methods at early stages, which leads to delayed treatment and poor prognosis. However, it is known that early detection and treatment of esophageal cancers can lead to long‐term survival [3].

Serum squamous cell carcinoma antigen (SCC‐Ag) and carcinoembryonic antigen (CEA) are globally used biomarkers for esophageal cancer. While they are effective as indicators of recurrence, the sensitivities of SCC‐Ag and CEA are 25.0% and 17.0% for preoperative detection, respectively, which limits their utility as diagnostic biomarkers [10, 11, 12]. Circulating tumor DNA (ctDNA), which has gained attention in recent years, is useful as a marker for predicting recurrence and prognosis in esophageal cancer [13, 14, 15]. However, its sensitivity for early‐stage diagnosis (Stage 0–II) is not high enough at 58.8% to be applied as an early diagnostic biomarker [14]. Therefore, novel promising biomarkers have to be discovered for early detection of esophageal cancers, which is expected to improve the poor diagnosis of esophageal cancers and establish better therapeutic strategies.

Cancer‐related biomarkers often reflect systemic alterations caused by tumor growth, in addition to factors directly derived from cancer tissues [16, 17]. It is generally accepted that cancer‐derived factors are not necessarily more sensitive or detectable than those originating from systemic responses, particularly in early stages. In the context of clinical screening for esophageal cancer, it is crucial to identify more sensitive and specific biomarkers, regardless of their origin. In recent years, metabolome analysis using mass spectrometry has been a powerful method for evaluating metabolites in biological samples from cancer patients [18, 19]. Since less invasive testing is preferred for patient care, plasma and urine were highlighted for comprehensive metabolome analysis. In this study, we conducted metabolome analysis on plasma and urine from esophageal cancer patients, with emphasis on early stages, to explore novel promising biomarkers.

2. Materials and Methods

2.1. Study Design and Participants

This single‐center observational study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Tohoku University (approval nos. 2021‐1‐167 and 2024‐1‐770). Written informed consent or an opt‐out method (this was only possible when the sample was collected non‐invasively and non‐interventionally for retrospective assessment) was obtained from all participants. After prespecified exclusions (withdrawn consent, prior cancer history in controls), we analyzed plasma from 80 patients with esophageal cancer and 29 healthy individuals, urine from 59 patients and 61 healthy participants, and fecal DNA from 82 patients and 172 healthy participants (Tables S1–S6). The plasma, urine, and fecal cohorts partially overlapped. Among patients with esophageal cancer, 7 patients provided only plasma samples, 5 provided urine‐only samples, and 14 provided only fecal samples. In addition, 12 patients provided both plasma and urine samples, 26 provided both plasma and fecal samples, 7 provided both urine and fecal samples, and 35 provided all three specimen types. Among healthy individuals, 29 participants provided all three specimen types, and an additional 32 participants provided both urine and fecal samples. A further 111 fecal samples were obtained from an independent set of healthy participants. Age, sex, smoking, alcohol use, cancer stage, and body mass index (BMI) were recorded using standardized questionnaires and medical records. In this study, “early‐stage disease” and “early diagnosis” refer specifically to Stage I/II esophageal cancer and do not include carcinoma in situ.

2.2. Sample Collection

All plasma (EDTA plasma), urine, and fecal samples from patients with esophageal cancer were collected before the initiation of any anticancer treatment. All specimens were processed within 2 h of collection, aliquoted, and stored at −80°C until analysis. Urinary creatinine levels were measured for normalization. Fecal samples were self‐collected using a stabilization buffer and stored at −80°C until subsequent DNA extraction.

Paired tumor and macroscopically non‐neoplastic esophageal tissues were obtained from 8 patients who underwent surgical resection without preoperative treatment. The macroscopically non‐neoplastic tissues were obtained from areas sufficiently distant from the tumor, and all surgical margins were pathologically negative. The tissue samples used for metabolomic analysis were not independently confirmed histologically to be non‐neoplastic.

2.3. Metabolite Extraction

Plasma and urine metabolites were extracted by protein precipitation and filtration. In brief, samples were mixed with methanol containing internal standards, phase‐separated (for plasma) or directly filtered (for urine) through a 5‐kDa cutoff, dried, and reconstituted with reference compounds immediately before analysis. Detailed volumes, internal standards, and workflow steps are provided in Doc S1.

2.4. Metabolome Analysis by CE‐TOFMS

CE‐TOFMS was performed using an Agilent CE‐TOFMS system with platinum ESI needle for anionic mode. Data were acquired with MassHunter and analyzed using MasterHands [20]. In cationic mode, metabolites were separated in a fused silica capillary (1 M formic acid, 30 kV) [21, 22]. The sheath liquid was 50% methanol with Hexakis. ESI‐TOFMS was run in positive mode with 4 kV and real‐time mass calibration ([2MeOH + H]+, Hexakis). In anionic mode, metabolites were separated using a COSMO(+) capillary with ammonium acetate buffer (pH 8.5, −30 kV). ESI‐TOFMS was run in negative mode with 3.5 kV, using acetate‐related lock masses for calibration [20, 23]. Detailed information can be found in Doc S1.

2.5. Statistical Analysis of Metabolomic Data

Among the 510 metabolites measured by the CE‐TOFMS, undetectable metabolites in either all samples or more than half of the samples derived from both esophageal cancer patients and healthy individuals were excluded. To adjust for potential confounders, a general linear model (GLM) was applied to metabolomic data to remove the effects of age, gender, smoking status, alcohol consumption, cancer stage, and body mass index. The adjusted p‐values were used to identify metabolites with statistically significant differences between esophageal cancer patients and healthy controls. Metabolites with false discovery rate (FDR) < 0.05 were considered significant. The general linear model analysis was conducted using MetaboAnalyst 5.0 (https://www.metaboanalyst.ca/) [24, 25]. Heatmaps based on Z‐scores were generated using Prism software to visualize metabolite distribution patterns.

Creatinine was included among the metabolites quantified by CE‐TOFMS. To assess the potential confounding effect of renal function on 3‐IS levels, plasma creatinine concentrations were compared between healthy individuals and patients with esophageal cancer.

Receiver operating characteristic (ROC) curves were constructed to evaluate the discriminatory ability of each metabolite. The area under the ROC curve (AUC) was calculated as a measure of diagnostic performance. To assess the internal validity and minimize the risk of overfitting, ROC analyses were internally validated using a bootstrap resampling method (1000 iterations). Bias‐corrected and accelerated 95% confidence intervals for the AUC were calculated. For the combined plasma and urine 3‐IS analysis, only participants with available measurements in both plasma and urine samples were utilized. A binary logistic regression model was fitted using plasma and urinary 3‐IS concentrations as continuous predictor variables and disease status as the binary outcome. The predicted probabilities generated by the model were used to construct the combined ROC curve.

To adjust for potential confounding factors, a general linear model was applied to the metabolomic data. Disease status, age, sex, smoking status, alcohol consumption, body mass index, and cancer stage were included in the model. For analytical purposes, healthy individuals were assigned a stage value of 0, whereas patients with esophageal cancer were assigned values of 1–4 according to their clinical stage. Stage was treated as an ordinal covariate to account for potential stage‐related variation in metabolite levels. Because cancer stage is not clinically defined for healthy individuals, an additional sensitivity analysis was performed after excluding stage from the model. The sensitivity model included disease status, age, sex, smoking status, alcohol consumption, and body mass index as covariates.

2.6. Targeted LC‐ESI‐MS/MS

Samples were processed by methanol extraction, dilution, and filtration as provided in Doc S2. Metabolites were spiked with 100 nM internal standards (pyridoxal‐5′‐phosphate‐d3 for 3‐IS, L‐valine‐13C5 for O‐AC, Bis‐S‐HPE‐AM‐d4 for indole) and quantified by LC‐ESI‐MS/MS (LCMS‐8060, SHIMADZU) coupled to Nexera UHPLC [26]. External calibration curves and blank controls were run each batch. Chromatographic conditions (columns, gradients, temperatures) and MRM transitions are provided in Doc S2.

2.7. 16S rRNA Gene Sequencing

Fecal DNA was extracted using a previously described bead‐beating method [27]. The V1‐V2 region of the 16S rRNA gene was amplified (about 10 ng DNA per reaction) with primers 27Fmod (5′‐AGRGTTTGATYMTGGCTCAG‐3′) and 338R (5′‐TGCTGCCTCCCGTAGGAGT‐3′), each containing overhang adapters [28]. PCR was performed with Tks Gflex DNA Polymerase (Takara) under the following program: 98°C for 1 min; 20 cycles of 98°C for 10 s, 55°C for 15 s, 68°C for 30 s; final extension at 68°C for 3 min. Libraries were sequenced using MiSeq (Illumina).

Sequence data were processed using QIIME 2 (version 2022.8) with DADA2 for quality filtering and denoising. For quality filtering, 20 and 19 bases were trimmed from the 5′ ends of the forward and reverse reads, respectively, and the forward and reverse reads were truncated at 280 and 210 bases, respectively. These truncation positions were selected so that the retained regions had approximate quality scores of ≥ 25. All other DADA2 parameters were left at their default settings. Amplicon sequence variants (ASVs) were inferred using DADA2, and taxonomic assignment was performed using the SILVA 16S rRNA reference database (version 132). Detected taxa are listed in Tables S5 and S6.

2.8. Cell Culture

Human esophageal cancer cell lines (KYSE70/150/170, ECGI‐10) were maintained in RPMI‐1640 with 10% FBS; immortalized esophageal epithelial cells (EPC2‐hTERT) were grown in keratinocyte serum‐free medium with supplements. Cells were cultured at 37°C, 5% CO2. EPC2‐hTERT was provided by Dr. Nakagawa (Columbia University) [29]. Detailed information can be found in Doc S3.

2.9. 3‐Indoxyl Sulfate Treatment

Human esophageal cancer cell lines and EPC2‐hTERT cells were seeded in 6‐well plates (1 × 105 cells/well) and cultured for 24 h. Cells were then treated with 0, 1, 100, or 250 μM of 3‐IS for 18 h. Omeprazole (250 μM) served as a positive control for AHR activation, and DMSO was used as vehicle control.

2.10. Quantitative Real‐Time PCR (qRT‐PCR)

Total RNA was extracted from cultured cells using ISOGEN (Nippon Gene), and reverse transcription into cDNA was performed using the ReverTra Ace qPCR RT Master Mix (TOYOBO) according to the manufacturer's protocol. qRT‐PCR was conducted using primers and THUNDERBIRD SYBR qPCR Mix (TOYOBO). β‐actin was used as an internal control for relative quantification among samples. The primers used are listed in Table S7.

2.11. Statistical Analysis

Quantitative data are presented as the means ± SD. For bivariate analyses, the normality was checked by using the Shapiro–Wilk test. For normally and non‐normally distributed data, Student's t‐test and Mann–Whitney U test were used to evaluate significant differences, respectively. For chronological analyses of culture media, multivariate analysis was performed using one‐way ANOVA. For paired comparisons between tumor and macroscopically non‐neoplastic tissues, the Wilcoxon matched‐pairs signed‐rank test was used. Statistical significance was indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

3. Results

3.1. Elevated 3‐Indoxyl Sulfate and O‐Acetylcarnitine in Plasma of Esophageal Cancer Patients

To discover candidates of promising novel biomarkers in esophageal cancer patients, we collected plasma samples from 80 esophageal cancer patients and 29 healthy individuals, and urine samples from 59 esophageal cancer patients and 61 healthy individuals. Using capillary electrophoresis‐time of flight mass spectrometry (CE‐TOFMS), we measured 510 metabolites (Tables S1–S4; Figures 1A and 2A). The data from the plasma samples were analyzed by a general linear model analysis removing 6 confounding factors: age, sex, smoking, alcohol consumption, cancer stage, and BMI. To examine the influence of stage coding on the results, we performed an additional sensitivity analysis using a general linear model that excluded cancer stage and included age, sex, smoking, alcohol consumption, and BMI as covariates.

FIGURE 1.

FIGURE 1

Plasma metabolome profiling of esophageal cancer patients and healthy individuals. (A) Plasma collection and CE‐TOFMS workflow. (B) p‐values with and without covariate adjustment (general linear model). (C) Heatmap of significantly altered metabolites. (D) Elevated plasma metabolites (O‐AC, dodecanoate, 3‐IS, uridine). Unless otherwise indicated, data are presented as mean ± SD. Statistical analyses were performed as described in the Methods. **p < 0.01; ****p < 0.0001.

FIGURE 2.

FIGURE 2

Urinary metabolome profiling of esophageal cancer patients and healthy individuals. (A) Urine collection and CE‐TOFMS workflow. (B) p‐values with and without covariate adjustment (general linear model). (C) Heatmap of differential metabolites. (D) Elevated urinary metabolites including 3‐IS, taurocyamine, and others. *p < 0.05; **p < 0.01; ****p < 0.0001.

The p‐values were calculated after covariate adjustment and plotted against those obtained without covariate adjustment (Figure 1B). As a result, 4 elevated metabolites and 18 decreased metabolites were identified in esophageal cancer patients compared to healthy individuals (Figure 1C). The decreased metabolites involved many amino acid‐related metabolites, which are considered to be due to reduced food intake [30], because esophageal cancer patients often exhibit dysphagia. Additionally, other factors unrelated to cancers, such as exercise, stress, and diet, can also affect these levels, even though serum amino acid levels could be cancer biomarkers [31]. Therefore, in this study, we focused on the elevated metabolites in the plasma of esophageal cancer patients, which were O‐acetylcarnitine (O‐AC), dodecanoate, 3‐indoxyl sulfate (3‐IS), and uridine (Figure 1D). Of particular note, in the sensitivity analysis, although the adjusted p‐values differed from those obtained in the original model, both 3‐IS and O‐AC remained significantly elevated in patients with esophageal cancer, and the direction of the associations was unchanged.

Urine samples were also analyzed by CE‐TOFMS, and the metabolite levels were normalized by creatinine levels, which were comparable between esophageal cancer patients and healthy individuals (Figure 2A). Removing the influence of the confounding factors, we calculated p‐values with covariate adjustment, which were plotted against those obtained without covariate adjustment (Figure 2B), resulting in the identification of 7 elevated metabolites and 28 decreased metabolites in esophageal cancer patients compared to healthy individuals. The 7 elevated metabolites were 3‐IS, taurocyamine, N1‐acetylspermidine, 5‐Methylthioadenosine, hippurate, kynurenine, and guanosine (Figure 2C).

Comparing the results of plasma and urine metabolites, 3‐IS was the only metabolite increased in both plasma and urine of esophageal cancer patients. Although no significant difference was observed for O‐AC in urine, it showed the most significant difference in plasma. Because 3‐IS is a known uremic toxin and its circulating levels can be influenced by renal function, we additionally compared plasma creatinine concentrations between healthy individuals and patients with esophageal cancer. Plasma creatinine concentrations were comparable between healthy individuals and patients with esophageal cancer (Figure S3). Thus, we focused on 3‐IS and O‐AC in the following sections.

Because significant differences in age, smoking status, and alcohol consumption were observed between patients with esophageal cancer and healthy controls, we conducted stratified analyses of 3‐IS and O‐AC according to these potential confounding factors (Figure S1). Elevated O‐AC and 3‐IS levels were observed in both the SCC and adenocarcinoma subgroups of patients with esophageal cancer (Figure S1A). Although no statistically significant differences were detected between the two histological subtypes, the cohort predominantly consisted of patients with SCC and included only a limited number of adenocarcinoma cases. Therefore, these subgroup analyses were exploratory and were not sufficiently powered to determine whether biomarker levels differ between SCC and adenocarcinoma. Similarly, no significant differences were observed among patients stratified by smoking status, alcohol consumption, or age group. These findings suggest that the associations of O‐AC and 3‐IS with esophageal cancer were generally consistent across the examined subgroups; however, given the observational nature of the study and limited sample size in some subgroups, residual confounding cannot be completely excluded.

3.2. Stability and Reliability of 3‐Indoxyl Sulfate and O‐Acetylcarnitine as Biomarkers

In order to utilize the metabolites as biomarkers, their stability and technical reproducibility in detection are essential and highly affect the reliability of the biomarkers. We examined the reproducibility of the results on 3‐IS and O‐AC detected by CE‐TOFMS by using a different mass spectrometry, liquid chromatography–tandem mass spectrometry (LC–MS/MS) (Figure 3A). Standard chemicals of 3‐IS and O‐AC were successfully detected by LC–MS/MS and exhibited resistance to repeated freeze–thaw procedures (data not shown). Using the established technique, we examined 3‐IS and O‐AC levels in the identical plasma and urine samples used in Figures 1 and 2, which were subjected to at least one more freeze–thaw cycle. The results of 3‐IS and O‐AC reproduced those detected by CE‐TOFMS in Figures 1D and 2D, with an improvement in 3‐IS detection as shown by no undetectable plasma samples (Figure 3B). These results indicate the good stability of 3‐IS and O‐AC in plasma and urine samples and suggest that both 3‐IS and O‐AC likely possess the potential for good biomarkers.

FIGURE 3.

FIGURE 3

Stability and reproducibility confirmed in blood and urine samples from esophageal cancer patients and healthy individuals. (A) Chromatograms of 3‐IS (100 nM) and O‐AC (100 nM) by LC–MS/MS. (B) Quantification results of blood and urine samples by LC–MS/MS following their use in CE‐TOF measurements and undergoing at least one freeze–thaw cycle. **p < 0.01; ****p < 0.0001.

3.3. Comparison of 3‐Indoxyl Sulfate and O‐Acetylcarnitine Levels Between Esophageal Cancer and Macroscopically Non‐Neoplastic Tissues

To examine the relationship of 3‐IS and O‐AC with esophageal cancer tissues, we measured these metabolites in paired tumor and macroscopically non‐neoplastic tissues obtained from eight patients who had not received preoperative treatment (Figure 4A). 3‐IS levels were significantly higher in esophageal cancer tissues than in macroscopically non‐neoplastic tissues, although one patient showed a relatively high 3‐IS level in the macroscopically non‐neoplastic tissues (Figure 4B). O‐AC levels tended to be higher in esophageal cancer tissues, but the difference did not reach statistical significance (p = 0.0547; Figure 4C). These findings demonstrate significantly increased 3‐IS levels, whereas O‐AC showed only a trend toward higher levels in esophageal cancer tissues. We subsequently investigated the cellular handling of these metabolites in vitro.

FIGURE 4.

FIGURE 4

Comparison of 3‐indoxyl sulfate and O‐acetylcarnitine between macroscopically non‐neoplastic and esophageal cancer tissues from the same patients. (A) Sampling area of esophageal cancer and macroscopically non‐neoplastic tissues. (B) 3‐IS concentrations in paired tumor and macroscopically non‐neoplastic tissues from 8 patients. (C) O‐AC concentrations in paired tumor and macroscopically non‐neoplastic tissues. O‐AC showed a trend toward higher levels in tumor tissues, although the difference did not reach statistical significance (p = 0.0547). Left panels show individual paired values, and right panels show the overall concentrations. Statistical comparisons were performed using the Wilcoxon matched‐pairs signed‐rank test. *p < 0.05. NT, macroscopically non‐neoplastic tissue.

3.4. 3‐Indoxyl Sulfate Activates Aryl Hydrocarbon Receptor in Esophageal Cancer Cells

3‐IS is a derivative of tryptophan, produced through the combined actions of commensal bacteria and the hepatic detoxification pathway [32, 33]. Higher levels of 3‐IS in esophageal cancer tissues raised a question whether esophageal cancer cells produce 3‐IS intracellularly or uptake it from extracellular sources (see Figure 4B). To answer this question, we cultured esophageal cancer cell lines and an immortalized normal human esophageal epithelial cell line with and without 3‐IS and measured its intracellular concentrations. In the absence of 3‐IS treatment, 3‐IS was not detected in the esophageal cancer cell lines or a normal cell line, indicating that 3‐IS is not intracellularly produced (Figure 5A), which is consistent with a previous study [32, 33]. In contrast, when 3‐IS was added to the culture medium, the intracellular 3‐IS amount was increased in a dose‐dependent manner in both esophageal cancer cell lines and the normal cell line (Figure 5A). These results indicate that esophageal cancer cells do not produce 3‐IS intracellularly, but instead take it up from the extracellular environment.

FIGURE 5.

FIGURE 5

Uptake of 3‐indoxyl sulfate and activation of the aryl hydrocarbon receptor (AHR) in esophageal cancer cell lines and an immortalized normal human esophageal epithelial cell line. (A) Dose‐dependent uptake of 3‐IS in esophageal cancer cell lines (KYSE 70, KYSE 150, KYSE 170, ECGI‐10) and an immortalized normal human esophageal epithelial cell line (EPC2‐hTERT). (B, C) Expression of AHR target genes (Cyp1A1, Cyp1A2) after 3‐IS treatment in each cell line. OME, omeprazole (250 μM). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

It has been reported that 3‐IS acts as a ligand for the aryl hydrocarbon receptor (AHR) [34, 35]. AHR is a ligand‐activated transcription factor and plays an important role in metabolizing xenobiotic substances by enhancing expression of detoxification enzymes such as CYP1A1 and CYP1A2 [36, 37]. In both esophageal cancer cell lines and the normal cell line, 3‐IS treatment increased expression of CYP1A1 and CYP1A2 in a dose‐dependent manner (Figure 5B,C). These suggest that, upon exposure to 3‐IS, both cancer cells and normal cells have the capacity to induce AHR activation.

3.5. Prevalence of Indole‐Producing Bacteria in Esophageal Cancer Patients

An initial step of 3‐IS production has been shown to be carried out by indole‐producing bacteria in gut microbiota [38]. Tryptophan is converted to indole by gut indole‐producing bacteria [39, 40], and the indole is released into the bloodstream and converted to 3‐IS in the liver [32, 33, 41]. Since 3‐IS levels were elevated in the plasma and urine of esophageal cancer patients, but esophageal cancer cells do not produce 3‐IS, we hypothesized that indole‐producing bacteria might be predominant in their gut. Twenty‐nine types of indole‐producing bacteria have been reported at the genus level [42]. We conducted fecal 16S rRNA analysis of 82 esophageal cancer patients and 172 healthy individuals (Tables S5 and S6) and found an increase in indole‐producing bacteria at the genus level in esophageal cancer patients (Figure 6A). When stratified by histological subtype, the putative indole‐producing bacteria similarly tended to be increased in both SCC and adenocarcinoma cases, although statistical power was limited due to the small number of adenocarcinoma patients (n = 12) (Figure S2). In particular, Fusobacterium, Enterobacter, and Escherichia‐Shigella were significantly altered in esophageal cancer patients (Figure 6B). Consistently, indole was increased in the plasma of esophageal cancer patients (Figure 6C).

FIGURE 6.

FIGURE 6

Comparison of the relative abundance of putative indole‐producing genera in fecal samples. (A) Relative abundance of genera previously reported to include indole‐producing bacterial species. (B) Putative indole‐producing genera showing significant differences between the two groups. (C) Representative indole chromatogram (100 nM) and plasma indole concentrations between the two groups. *p < 0.05; ***p < 0.001.

Although, due to the genus‐level resolution of 16S rRNA analysis, functional interpretation should be made with caution, these results suggest a positive association between increased 3‐IS levels and the relative abundance of indole‐producing genera in the gut of esophageal cancer patients.

3.6. O‐Acetylcarnitine Released From Esophageal Cancer Cells

O‐AC is an acetylated form of L‐carnitine, which is generated from acetyl‐CoA in mitochondria as a consequence of glucose and fatty acid catabolism [43, 44]. Consistently, O‐AC was present in all cell lines, and esophageal cancer cell lines tended to produce it more abundantly than the immortalized normal cell lines (Figure 7A). To investigate whether O‐AC is secreted from these cell lines, O‐AC concentration in culture media was measured at the indicated time points. Intriguingly, O‐AC was increased in the culture supernatant of esophageal cancer cell lines after 72 h of culture, whereas O‐AC was detected at low levels in the culture supernatant of a normal cell line (Figure 7B), suggesting that esophageal cancer cells may contribute to extracellular O‐AC levels under in vitro conditions.

FIGURE 7.

FIGURE 7

O‐acetylcarnitine in esophageal cancer cell lines and an immortalized normal human esophageal epithelial cell line. (A) Intracellular O‐AC in esophageal cancer versus normal cells. (B) Time‐dependent O‐AC accumulation in supernatants. *p < 0.05; **p < 0.01.

3.7. 3‐Indoxyl Sulfate and O‐Acetylcarnitine as Promising Early‐Stage Biomarkers in Esophageal Cancer

To further evaluate the potential of 3‐IS and O‐AC as early biomarkers for esophageal cancer, 3‐IS in plasma and urine, along with O‐AC in plasma, were compared across different stages of the disease (Figure 8A). 3‐IS tended to be increased in both plasma and urine across all stages, with an increasing trend in the early stages, Stage I/II. O‐AC was also significantly upregulated across all stages in plasma. To assess the potential of 3‐IS and O‐AC as biomarkers of esophageal cancers, sensitivity, specificity, and AUC were evaluated by creating ROC curves (Figure 8B). The sensitivity, specificity, and AUC of 3‐IS in plasma were 0.72, 0.80, and 0.791 for Stage I/II (Figure 8B), whereas those were slightly lower for all stages (Figure 9A). In urine, 3‐IS showed similar results with a sensitivity of 0.79, a specificity of 0.67, and an AUC of 0.777 for Stage I/II (Figures 8B and 9A). To further evaluate the combined diagnostic performance of plasma and urinary 3‐IS, we performed a ROC analysis using paired samples. Among participants with paired plasma and urine samples, the combined 3‐IS ROC analysis included 17 patients with Stage I/II esophageal cancer and 29 healthy individuals. A binary logistic regression model incorporating plasma and urinary 3‐IS concentrations yielded an AUC of 0.863 (Figure 8C). To assess the internal validity and minimize the risk of overfitting, bootstrap resampling (1000 iterations) was performed. In Stage I/II, the combined plasma and urine 3‐IS model showed an AUC of 0.87 (95% CI, 0.76–0.96), suggesting that examining both sample types can enhance the utility of 3‐IS as a promising biomarker.

FIGURE 8.

FIGURE 8

Biomarker utility of 3‐indoxyl sulfate in plasma and urine and O‐acetylcarnitine in plasma. (A) 3‐IS and O‐AC levels by disease stage. (B) ROC curves for plasma and urine 3‐IS (Stage I/II). (C) Multivariate plasm + urine 3‐IS ROC (AUC = 0.863)" with "ROC curve for combined plasma and urine 3‐IS (AUC = 0.863). (D) ROC curve for plasma O‐AC (Stage I/II). *p < 0.05; **p < 0.01; ****p < 0.001; ****p < 0.0001.

FIGURE 9.

FIGURE 9

Biomarker utility of 3‐indoxyl sulfate in plasma and urine, and O‐acetylcarnitine in plasma at all stages. (A) ROC curves for plasma and urine 3‐IS across all stages. (B) ROC curve for plasma O‐AC across all stages.

On the other hand, O‐AC in plasma showed a sensitivity of 0.85, a specificity of 0.90, and an AUC of 0.916 across all stages (Figure 9B) and with 0.83, 0.87, and 0.902 for Stage I/II (Figure 8D). Bootstrap internal validation yielded the AUC of 0.91 (95% CI, 0.85–0.96) and 0.90 (95% CI, 0.82–0.97) in all stages and Stage I/II, respectively, suggesting the potential utility of O‐AC as a biomarker for early detection.

4. Discussion

In this study, we conducted comprehensive metabolome analysis on plasma and urine derived from esophageal cancer patients and identified 3‐IS and O‐AC as promising candidate biomarkers for the early diagnosis of esophageal cancer. 3‐IS was concurrently upregulated in both plasma and urine of the patients, and the AUC of multivariate ROC curves of plasma and urine samples in Stage I/II was 0.863 for 3‐IS. In contrast, O‐AC was the most significantly elevated in plasma of the patients while not significant in urine. ROC curve in Stage I/II showed a high AUC of 0.902. These results suggest that 3‐IS and O‐AC alone or their combination are potential biomarkers for early diagnosis.

Although 3‐IS was elevated in both plasma and urine of esophageal cancer patients, it did not seem to be intracellularly produced in esophageal cancer cells. 3‐IS has been known as a metabolite of tryptophan [32, 33]. Tryptophan is converted into indole by gut bacteria that produce tryptophanase in the intestine [39, 40], circulates in the bloodstream and is converted into 3‐IS in liver by the hepatic enzymes, CYP2E1 and SULT1A1, after which 3‐IS is finally excreted into urine [32, 33, 41]. We found that Fusobacterium was increased in fecal samples from esophageal cancer patients (see Figure 6B). The increased fecal indole‐producing bacteria accompanied by elevated indole in the plasma of esophageal cancer patients (see Figure 6C) suggest that Fusobacterium may be associated with the metabolic environment of esophageal cancer.

A recent study has shown that an indole‐producing bacterium, Fusobacterium, invades esophageal cancer tissues and contributes to the development and progression of esophageal cancer by evoking inflammation in the cancer tissues [45, 46]. We also observed that Fusobacterium is localized in esophageal cancer tissues [47]. Thus, Fusobacterium may be associated with the development and propagation of esophageal cancer through multiple mechanisms. However, as a limitation of the current study, the contribution of Fusobacterium to indole production cannot be determined from the present data. Furthermore, genus‐level 16S rRNA sequencing does not allow determination of the functional capacity of individual species or strains, nor does it permit reliable functional or causal inference. Therefore, the observed microbiome alterations should be interpreted with caution [45]. How the Fusobacterium expands in fecal samples of esophageal cancer patients requires further investigation.

It has been reported that 3‐IS is taken up by cancer cells and activates AHR [34, 35]. AHR activation in cancers has been reported to promote cell proliferation and anti‐apoptotic effects, as well as the formation of an immunosuppressive environment, all of which are associated with cancer progression [48, 49]. In particular, AHR expression is negatively correlated with the efficacy of anticancer drug treatment [48]. We found that 3‐IS concentrations in cancer tissues were higher than in macroscopically non‐neoplastic tissues, but much lower compared to cancer cell lines (Figures 4B and 5A). Since cancer tissues contain extracellular proteins, which are included in the total protein amount used for normalization, this may lead to an underestimation of 3‐IS concentrations in cancer tissues compared with cultured cell lines. In the in vitro experiments, the uptake of 3‐IS and induction of AHR target genes were observed in both esophageal cancer cell lines and the non‐neoplastic EPC2‐hTERT cell line. Therefore, although 3‐IS is enriched in esophageal cancer tissues, the biological significance of 3‐IS‐mediated AHR signaling might not be specific for cancer tissues, as 3‐IS was also increased in plasma.

O‐AC was elevated in plasma of esophageal cancer patients and also tended to increase in esophageal cancer tissues. Our in vitro experiments showed that esophageal cancer cells can produce and secrete this metabolite. O‐AC is closely related to the mitochondrial metabolism; O‐AC is produced in mitochondria matrix by carnitine O‐acetyltransferase (CrAT), which transfers acetyl groups from Acetyl‐CoA to carnitine, and the O‐AC translocates to cytoplasm, transfer the acetyl group to cytoplasmic CoA, and thereby acts as an acetyl group supplier [44]. The supplied acetyl groups are used for various cellular events, i.e., for epigenetic regulation. While the function of O‐AC in esophageal cancers has not been well examined, alteration of carnitine metabolism has been recognized as hallmarks of cancers and promotes the metabolic plasticity of cancers [50, 51]. Therefore, the elevated O‐AC may reflect the altered metabolic processes associated with esophageal cancer progression, although its direct role remains unclear.

Esophageal cancer patients can achieve long‐term survival if diagnosed at early stages [3]. To this end, many efforts have been recently made for finding new early biomarkers, in addition to traditional conventional blood biomarkers for esophageal cancers, like SCC‐Ag, CEA, and ctDNA. Our results suggest that 3‐IS and O‐AC are potential biomarkers for early‐stage esophageal cancer, which warrants further investigation. Furthermore, examining 3‐IS and O‐AC can be conducted with low‐invasive blood and urine tests, which also holds significant clinical importance. In addition, it has recently been reported that miRNAs can be useful blood‐based biomarkers for esophageal cancer diagnosis [52, 53, 54]. A multivariate ROC curve using the miRNAs showed an AUC of over 0.9 for the early stage (Stage I/II) of esophageal cancer. Combination of these miRNAs and our candidates for early diagnosis may establish more accurate and reliable diagnostic criteria for early‐stage esophageal cancers.

However, several limitations of this study should be acknowledged. First, the current study was conducted using a single‐center cohort with a relatively limited sample size, and the control group consisted exclusively of healthy individuals. Accordingly, although 3‐IS and O‐AC showed good discriminatory performance in distinguishing patients with esophageal cancer and healthy individuals, their disease specificity remains unclear, and these metabolites may reflect systemic metabolic alterations rather than tumor‐specific changes [50]. Future studies including patients with benign esophageal disorders, inflammatory diseases, and other malignancies are warranted to determine whether these metabolites are specific biomarkers for esophageal cancer. In addition, the clinical cohorts predominantly consisted of patients with SCC, and the number of adenocarcinoma cases was limited. Consequently, the present study was not sufficiently powered to determine whether the diagnostic performance of 3‐IS and O‐AC differs between SCC and adenocarcinoma. Further validation in larger adenocarcinoma cohorts and functional studies using adenocarcinoma‐derived experimental models will be required to determine the generalizability of the present findings. Regarding the limitation in the current cohorts, the ROC analyses were also performed in the same single‐center cohort. Although bootstrap resampling supported the internal validity of the findings, independent external validation will be required before the clinical utility of these candidate biomarkers can be established. Second, although statistical adjustments and stratified analyses were performed, substantial differences in baseline characteristics between patients with esophageal cancer and healthy individuals, particularly in age, smoking status, and alcohol consumption, remain an important limitation of this study. Although these variables were included in the general linear models and were additionally examined in stratified analyses, residual confounding cannot be completely excluded. Third, although plasma creatinine concentrations did not significantly differ between the groups (Figure S3), residual confounding by renal function cannot be completely excluded, particularly for 3‐IS, and should be further evaluated using clinical renal function measures, such as estimated glomerular filtration rate and chronic kidney disease status. Fourth, microbiome analyses were based on genus‐level 16S rRNA sequencing, which restricts functional and causal interpretation, as aforementioned. Fifth, the origin of elevated 3‐IS has not been identified, and the specificity and functions of 3‐IS‐mediated AHR signaling in esophageal cancer tissues have not been elucidated in this study. Therefore, future multicenter studies with larger cohorts combined with in‐depth functional analyses will be essential to clarify the clinical and biological significance of the present findings.

The principal findings of this study are summarized schematically in Figure 10. In conclusion, 3‐IS and O‐AC may represent promising candidate biomarkers for the early detection of esophageal cancer. Further validation in independent cohorts, including disease‐control populations, is required before their clinical application.

FIGURE 10.

FIGURE 10

An illustration summarizing the key findings and significance of this study. NT, macroscopically non‐neoplastic tissue.

Author Contributions

Kenji Hamada: writing – original draft, investigation, formal analysis. Yohei Ozawa: supervision, funding acquisition, resources, project administration, writing – review and editing, conceptualization, methodology. Hiroshi Okamoto: supervision, resources. Shinji Fukuda: supervision, investigation. Tomoyoshi Soga: supervision, funding acquisition, investigation. Takaaki Abe: supervision. Hozumi Motohashi: supervision, writing – review and editing. Shohei Murakami: writing – review and editing, conceptualization, investigation, validation, methodology, supervision. Takashi Kamei: supervision, resources, writing – review and editing.

Funding

This work was supported by AMED Grant Number 25zf0127001h0005 (T.A.), JP21zf0127001 (T.S.), JST CREST Grant Number JPMJCR2123 (T.S.), MEXT KAKENHI Grant Numbers JP23H04946 (T.S.), JP22H03541 (S.F.), World Premier International Research Center Initiative (WPI), Human Biology‐Microbiome‐Quantum Research Center (Bio2Q) (T.S.), MEXT, and MEXT Promotion of Development of a Joint Usage/Research System Project: Coalition of Universities for Research Excellence Program (CURE) Grant Number JPMXP1323015484 (T.S.), MEXT, and JSPS Grant Numbers 24K11531 (S.M.), 23K08206 (Y.O.), 21H04799 (H.M.), 21H05264 (H.M.), 21H05258 (H.M.) and 24H00605 (H.M.). The funders had no role in study design, data collection, data analysis, or manuscript preparation.

Ethics Statement

This study was conducted using human biological samples obtained from patients, in accordance with the Declaration of Helsinki, and was approved by the Ethics Committee of Tohoku University (approval nos. 2021‐1‐167 and 2024‐1‐770).

Consent

Informed consent was obtained from all participants.

Conflicts of Interest

Hozumi Motohashi serves as an associate editor of Cancer Science. All other authors declare no conflicts of interest.

Supporting information

Doc S1: Metabolite extraction and CE‐TOFMS analysis.

Doc S2: LC‐ESI‐MS/MS analysis.

Doc S3: Cell culture and treatments.

CAS-9999-0-s001.docx (26.9KB, docx)

Figure S1: Stratified analyses of plasma 3‐indoxyl sulfate and O‐acetylcarnitine according to potential confounding factors. (A) Tissue origin, (B) smoking status, (C) alcohol consumption, and (D) age group. Data are presented as mean ± SD with individual data points. ND, never drinker; NS, never smoker; CD, current drinker; CS, current smoker. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure S2: Genus‐level composition of putative indole‐producing bacteria in fecal samples stratified by histological subtype. The relative abundance of putative indole‐producing genera based on genus‐level 16S rRNA sequencing was compared between SCC and adenocarcinoma groups. Due to the limited number of adenocarcinoma cases, these subgroup analyses should be interpreted with caution. **p < 0.01.

Figure S3: Comparison of plasma creatinine concentrations between healthy individuals and patients with esophageal cancer. Plasma creatinine concentrations quantified by CE‐TOFMS were compared between healthy individuals (n = 29) and patients with esophageal cancer (n = 80). Data are presented as mean ± SD. No significant difference was observed between the two groups (p = 0.58).

CAS-9999-0-s003.docx (592.6KB, docx)

Table S1: Clinical characteristics of participants providing plasma samples.

Table S2: Metabolite concentrations in plasma samples.

Table S3: Clinical characteristics of participants providing urine samples.

Table S4: Metabolite concentrations in urine samples.

Table S5: Clinical characteristics of participants providing fecal samples.

Table S6: Bacterial taxon abundances in fecal samples.

Table S7: qPCR primers.

CAS-9999-0-s002.zip (112.4KB, zip)

Acknowledgments

This study was conducted using the shared equipment facility at the Institute of Development, Aging and Cancer, Tohoku University.  We sincerely thank the staff of the facility for their technical support.

Contributor Information

Yohei Ozawa, Email: yohei.ozawa.c1@tohoku.ac.jp.

Shohei Murakami, Email: shohei.murakami.a6@tohoku.ac.jp.

Data Availability Statement

Research data are not shared.

References

  • 1. Sung H., Ferlay J., Siegel R. L., et al., “Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: A Cancer Journal for Clinicians 71, no. 3 (2021): 209–249, 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
  • 2. Xu J. Q., Zhang Z. C., Chen W. F., et al., “Repeat Endoscopic Submucosal Dissection as Salvage Treatment for Local Recurrence of Esophageal Squamous Cell Carcinoma After Initial Endoscopic Submucosal Dissection,” Gastrointestinal Endoscopy 96, no. 1 (2022): 18–27.e1, 10.1016/j.gie.2022.02.043. [DOI] [PubMed] [Google Scholar]
  • 3. Cancer Statistics (Cancer Information Service, National Cancer Center, Vital Statistics of Japan, Ministry of Health, Labour and Welfare, 2025), https://ganjoho.jp/reg_stat/statistics/data/dl/en.html. [Google Scholar]
  • 4. Miyata H., Yamasaki M., Kurokawa Y., et al., “Survival Factors in Patients With Recurrence After Curative Resection of Esophageal Squamous Cell Carcinomas,” Annals of Surgical Oncology 18, no. 12 (2011): 3353–3361, 10.1245/s10434-011-1747-7. [DOI] [PubMed] [Google Scholar]
  • 5. Lou F., Sima C. S., Adusumilli P. S., et al., “Esophageal Cancer Recurrence Patterns and Implications for Surveillance,” Journal of Thoracic Oncology 8, no. 12 (2013): 1558–1562, 10.1097/01.JTO.0000437420.38972.fb. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Hamai Y., Hihara J., Emi M., et al., “Treatment Outcomes and Prognostic Factors After Recurrence of Esophageal Squamous Cell Carcinoma,” World Journal of Surgery 42, no. 7 (2018): 1, 10.1007/s00268-017-4430-8. [DOI] [PubMed] [Google Scholar]
  • 7. Arnal M. J. D., Arenas Á. F., and Arbeloa Á. L., “Esophageal Cancer: Risk Factors, Screening and Endoscopic Treatment in Western and Eastern Countries,” World Journal of Gastroenterology 21, no. 26 (2015): 7933–7943, 10.3748/wjg.v21.i26.7933. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Lauby‐Secretan B., Scoccianti C., Loomis D., Grosse Y., Bianchini F., and Straif K., “Body Fatness and Cancer—Viewpoint of the IARC Working Group,” New England Journal of Medicine 375, no. 8 (2016): 794–798, 10.1056/NEJMsr1606602. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Chen Z., Zhang X., Zhai J., et al., “Global Burden of Esophageal Cancer Attributable to High BMI in 204 Countries and Territories: 1990–2019,” Thorac Cancer 15, no. 9 (2024): 681–692, 10.1111/1759-7714.15239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Shimada H., Nabeya Y., Okazumi S. i., et al., “Prediction of Survival With Squamous Cell Carcinoma Antigen in Patients With Resectable Esophageal Squamous Cell Carcinoma,” Surgery 133, no. 5 (2003): 486–494, 10.1067/msy.2003.139. [DOI] [PubMed] [Google Scholar]
  • 11. Mroczko B., Kozłowski M., Groblewska M., et al., “The Diagnostic Value of the Measurement of Matrix Metalloproteinase 9 (MMP‐9), Squamous Cell Cancer Antigen (SCC) and Carcinoembryonic Antigen (CEA) in the Sera of Esophageal Cancer Patients,” Clinica Chimica Acta 389, no. 1 (2008): 61–66, 10.1016/j.cca.2007.11.023. [DOI] [PubMed] [Google Scholar]
  • 12. Banki F., Yacoub W. N., Hagen J. A., et al., “Plasma DNA Is More Reliable Than Carcinoembryonic Antigen for Diagnosis of Recurrent Esophageal Cancer,” Journal of the American College of Surgeons 207, no. 1 (2008): 30–35, 10.1016/j.jamcollsurg.2008.01.004. [DOI] [PubMed] [Google Scholar]
  • 13. Pessoa L. S., Heringer M., and Ferrer V. P., “ctDNA as a Cancer Biomarker: A Broad Overview,” Critical Reviews in Oncology/Hematology 155 (2020): 103109, 10.1016/j.critrevonc.2020.103109. [DOI] [PubMed] [Google Scholar]
  • 14. Qiao G., Zhuang W., Dong B., et al., “Discovery and Validation of Methylation Signatures in Circulating Cell‐Free DNA for Early Detection of Esophageal Cancer: A Case‐Control Study,” BMC Medicine 19, no. 1 (2021): 243, 10.1186/s12916-021-02109-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Zhang Y., Du H., Wang N., Wang L., and Huang Y., “An Update of Clinical Value of Circulating Tumor DNA in Esophageal Cancer: A Systematic Review and Meta‐Analysis,” BMC Cancer 24, no. 1 (2024): 129, 10.1186/s12885-024-11879-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Gao H., Dong B., Jia J., et al., “Application of Ex Vivo 1H NMR Metabonomics to the Characterization and Possible Detection of Renal Cell Carcinoma Metastases,” Journal of Cancer Research and Clinical Oncology 138, no. 5 (2012): 753–761, 10.1007/s00432-011-1134-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Wu J., Guo Z., and Gao Y., “Dynamic Changes of Urine Proteome in a Walker 256 Tumor‐Bearing Rat Model,” Cancer Medicine 6, no. 11 (2017): 2713–2722, 10.1002/cam4.1225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Sugimoto M., Wong D. T., Hirayama A., Soga T., and Tomita M., “Capillary Electrophoresis Mass Spectrometry‐Based Saliva Metabolomics Identified Oral, Breast and Pancreatic Cancer‐Specific Profiles,” Metabolomics 6, no. 1 (2010): 78–95, 10.1007/s11306-009-0178-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Harada S., Ohmomo H., Matsumoto M., et al., “Metabolomics Profiles Alterations in Cigarette Smokers and Heated Tobacco Product Users,” Journal of Epidemiology 34, no. 9 (2024): 403–410, 10.2188/jea.je20230170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Soga T., Igarashi K., Ito C., Mizobuchi K., Zimmermann H. P., and Tomita M., “Metabolomic Profiling of Anionic Metabolites by Capillary Electrophoresis Mass Spectrometry,” Analytical Chemistry 81, no. 15 (2009): 6165–6174, 10.1021/ac900675k. [DOI] [PubMed] [Google Scholar]
  • 21. Soga T., Ohashi Y., Ueno Y., Naraoka H., Tomita M., and Nishioka T., “Quantitative Metabolome Analysis Using Capillary Electrophoresis Mass Spectrometry,” Journal of Proteome Research 2, no. 5 (2003): 488–494, 10.1021/pr034020m. [DOI] [PubMed] [Google Scholar]
  • 22. Soga T., Baran R., Suematsu M., et al., “Differential Metabolomics Reveals Ophthalmic Acid as an Oxidative Stress Biomarker Indicating Hepatic Glutathione Consumption*,” Journal of Biological Chemistry 281, no. 24 (2006): 16768–16776, 10.1074/jbc.M601876200. [DOI] [PubMed] [Google Scholar]
  • 23. Satoh K., Yachida S., Sugimoto M., et al., “Global Metabolic Reprogramming of Colorectal Cancer Occurs at Adenoma Stage and Is Induced by MYC,” Proceedings of the National Academy of Sciences 114, no. 37 (2017): E7697–E7706, 10.1073/pnas.1710366114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Pang Z., Zhou G., Ewald J., et al., “Using MetaboAnalyst 5.0 for LC–HRMS Spectra Processing, Multi‐Omics Integration and Covariate Adjustment of Global Metabolomics Data,” Nature Protocols 17, no. 8 (2022): 1735–1761, 10.1038/s41596-022-00710-w. [DOI] [PubMed] [Google Scholar]
  • 25. Pang Z., Chong J., Zhou G., et al., “MetaboAnalyst 5.0: Narrowing the Gap Between Raw Spectra and Functional Insights,” Nucleic Acids Research 49, no. W1 (2021): W388–W396, 10.1093/nar/gkab382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Takata T., Jung M., Matsunaga T., et al., “Methods in Sulfide and Persulfide Research,” Nitric Oxide: Biology and Chemistry 116 (2021): 47–64, 10.1016/j.niox.2021.09.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Nagai M., Moriyama M., Ishii C., et al., “High Body Temperature Increases Gut Microbiota‐Dependent Host Resistance to Influenza A Virus and SARS‐CoV‐2 Infection,” Nature Communications 14, no. 1 (2023): 3863, 10.1038/s41467-023-39569-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Kim S. W., Suda W., Kim S., et al., “Robustness of Gut Microbiota of Healthy Adults in Response to Probiotic Intervention Revealed by High‐Throughput Pyrosequencing,” DNA Research 20, no. 3 (2013): 241–253, 10.1093/dnares/dst006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Harada H., Nakagawa H., Oyama K., et al., “Telomerase Induces Immortalization of Human Esophageal Keratinocytes Without p16INK4a Inactivation,” Molecular Cancer Research 1, no. 10 (2003): 729–738. [PubMed] [Google Scholar]
  • 30. Church D. D., Hirsch K. R., Park S., et al., “Essential Amino Acids and Protein Synthesis: Insights Into Maximizing the Muscle and Whole‐Body Response to Feeding,” Nutrients 12, no. 12 (2020): 3717, 10.3390/nu12123717. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Wiggins T., Kumar S., Markar S. R., Antonowicz S., and Hanna G. B., “Tyrosine, Phenylalanine, and Tryptophan in Gastroesophageal Malignancy: A Systematic Review,” Cancer Epidemiology, Biomarkers & Prevention 24, no. 1 (2015): 32–38, 10.1158/1055-9965.EPI-14-0980. [DOI] [PubMed] [Google Scholar]
  • 32. Kuo S. Z., Dettmer K., Annavajhala M. K., et al., “Associations Between Urinary 3‐Indoxyl Sulfate, a Gut Microbiome‐Derived Biomarker, and Patient Outcomes After Intensive Care Unit Admission,” Journal of Critical Care 63 (2021): 15–21, 10.1016/j.jcrc.2021.01.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Banoglu E., Jha G. G., and King R. S., “Hepatic Microsomal Metabolism of Indole to Indoxyl, a Precursor of Indoxyl Sulfate,” European Journal of Drug Metabolism and Pharmacokinetics 26, no. 4 (2001): 235–240. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Schroeder J. C., DiNatale B. C., Murray I. A., et al., “The Uremic Toxin 3‐Indoxyl Sulfate Is a Potent Endogenous Agonist for the Human Aryl Hydrocarbon Receptor,” Biochemistry 49, no. 2 (2010): 393–400, 10.1021/bi901786x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Kim H. Y., Kang Y. J., Kim D. H., et al., “Uremic Toxin Indoxyl Sulfate Induces Trained Immunity via the AhR‐Dependent Arachidonic Acid Pathway in End‐Stage Renal Disease (ESRD),” eLife 12 (2024): RP87316, 10.7554/eLife.87316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Vogel C. F. A., van Winkle L. S., Esser C., and Haarmann‐Stemmann T., “The Aryl Hydrocarbon Receptor as a Target of Environmental Stressors – Implications for Pollution Mediated Stress and Inflammatory Responses,” Redox Biology 34 (2020): 101530, 10.1016/j.redox.2020.101530. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Coelho N. R., Pimpão A. B., Correia M. J., et al., “Pharmacological Blockage of the AHR‐CYP1A1 Axis: A Call for In Vivo Evidence,” Journal of Molecular Medicine 100, no. 2 (2022): 215–243, 10.1007/s00109-021-02163-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Skye S. M. and Hazen S. L., “Microbial Modulation of a Uremic Toxin,” Cell Host & Microbe 20, no. 6 (2016): 691–692, 10.1016/j.chom.2016.11.005. [DOI] [PubMed] [Google Scholar]
  • 39. Tennoune N., Andriamihaja M., and Blachier F., “Production of Indole and Indole‐Related Compounds by the Intestinal Microbiota and Consequences for the Host: The Good, the Bad, and the Ugly,” Microorganisms 10, no. 5 (2022): 930, 10.3390/microorganisms10050930. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Bansal T., Alaniz R. C., Wood T. K., and Jayaraman A., “The Bacterial Signal Indole Increases Epithelial‐Cell Tight‐Junction Resistance and Attenuates Indicators of Inflammation,” Proceedings of the National Academy of Sciences 107, no. 1 (2010): 228–233, 10.1073/pnas.0906112107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Hou H., Horikawa M., Narita Y., et al., “Suppression of Indoxyl Sulfate Accumulation Reduces Renal Fibrosis in Sulfotransferase 1a1‐Deficient Mice,” International Journal of Molecular Sciences 24, no. 14 (2023): 11329, 10.3390/ijms241411329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Lee J. H. and Lee J., “Indole as an Intercellular Signal in Microbial Communities,” FEMS Microbiology Reviews 34, no. 4 (2010): 426–444, 10.1111/j.1574-6976.2009.00204.x. [DOI] [PubMed] [Google Scholar]
  • 43. Rosca M. G., Lemieux H., and Hoppel C. L., “Mitochondria in the Elderly: Is Acetylcarnitine a Rejuvenator?,” Advanced Drug Delivery Reviews 61, no. 14 (2009): 1332–1342, 10.1016/j.addr.2009.06.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Izzo L. T., Trefely S., Demetriadou C., et al., “Acetylcarnitine Shuttling Links Mitochondrial Metabolism to Histone Acetylation and Lipogenesis,” Science Advances 9, no. 18 (2023): eadf0115, 10.1126/sciadv.adf0115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Lei J., Xu F., Deng C., et al., “ Fusobacterium nucleatum Promotes the Early Occurrence of Esophageal Cancer Through Upregulation of IL‐32/PRTN3 Expression,” Cancer Science 114, no. 6 (2023): 2414–2428, 10.1111/cas.15787. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Stasiewicz M. and Karpiński T. M., “The Oral Microbiota and Its Role in Carcinogenesis,” Seminars in Cancer Biology 86 (2022): 633–642, 10.1016/j.semcancer.2021.11.002. [DOI] [PubMed] [Google Scholar]
  • 47. Sasaki‐Higashimoto I., Fujishima F., Ishida H., et al., “Histopathological Study of the Localization/Distribution of Fusobacterium nucleatum in Esophageal Cancer,” Pathology International 75, no. 2 (2025): 82–91, 10.1111/pin.13505. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. To K. K. W., Yu L., Liu S., Fu J., and Cho C. H., “Constitutive AhR Activation Leads to Concomitant ABCG2‐Mediated Multidrug Resistance in Cisplatin‐Resistant Esophageal Carcinoma Cells,” Molecular Carcinogenesis 51, no. 6 (2012): 449–464, 10.1002/mc.20810. [DOI] [PubMed] [Google Scholar]
  • 49. Rahmati M., Moghtaderi H., Mohammadi S., and Al‐Harrasi A., “Aryl Hydrocarbon Receptor Dynamics in Esophageal Squamous Cell Carcinoma: From Immune Modulation to Therapeutic Opportunities,” World Journal of Experimental Medicine 14, no. 3 (2024):96269, 10.5493/wjem.v14.i3.96269. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Melone M. A. B., Valentino A., Margarucci S., Galderisi U., Giordano A., and Peluso G., “The Carnitine System and Cancer Metabolic Plasticity,” Cell Death & Disease 9, no. 2 (2018): 1–12, 10.1038/s41419-018-0313-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51. Lu Y., Li N., Gao L., et al., “Acetylcarnitine Is a Candidate Diagnostic and Prognostic Biomarker of Hepatocellular Carcinoma,” Cancer Research 76, no. 10 (2016): 2912–2920, 10.1158/0008-5472.CAN-15-3199. [DOI] [PubMed] [Google Scholar]
  • 52. Sharma P. and Sharma R., “miRNA–mRNA Crosstalk in Esophageal Cancer: From Diagnosis to Therapy,” Critical Reviews in Oncology/Hematology 96, no. 3 (2015): 449–462, 10.1016/j.critrevonc.2015.07.002. [DOI] [PubMed] [Google Scholar]
  • 53. Zheng Y. J., Liang T. S., Wang J., et al., “MicroRNA‐155 Acts as a Diagnostic and Prognostic Biomarker for Oesophageal Squamous Cell Carcinoma,” Artificial Cells, Nanomedicine, and Biotechnology 48, no. 1 (2020): 977–982, 10.1080/21691401.2020.1773479. [DOI] [PubMed] [Google Scholar]
  • 54. Liu Z., Huang Y., Han Z., et al., “Exosome‐Mediated miR‐25/miR‐203 as a Potential Biomarker for Esophageal Squamous Cell Carcinoma: Improving Early Diagnosis and Revealing Malignancy,” Translational Cancer Research 10, no. 12 (2021): 5174–5182, 10.21037/tcr-21-1123. [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

Doc S1: Metabolite extraction and CE‐TOFMS analysis.

Doc S2: LC‐ESI‐MS/MS analysis.

Doc S3: Cell culture and treatments.

CAS-9999-0-s001.docx (26.9KB, docx)

Figure S1: Stratified analyses of plasma 3‐indoxyl sulfate and O‐acetylcarnitine according to potential confounding factors. (A) Tissue origin, (B) smoking status, (C) alcohol consumption, and (D) age group. Data are presented as mean ± SD with individual data points. ND, never drinker; NS, never smoker; CD, current drinker; CS, current smoker. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure S2: Genus‐level composition of putative indole‐producing bacteria in fecal samples stratified by histological subtype. The relative abundance of putative indole‐producing genera based on genus‐level 16S rRNA sequencing was compared between SCC and adenocarcinoma groups. Due to the limited number of adenocarcinoma cases, these subgroup analyses should be interpreted with caution. **p < 0.01.

Figure S3: Comparison of plasma creatinine concentrations between healthy individuals and patients with esophageal cancer. Plasma creatinine concentrations quantified by CE‐TOFMS were compared between healthy individuals (n = 29) and patients with esophageal cancer (n = 80). Data are presented as mean ± SD. No significant difference was observed between the two groups (p = 0.58).

CAS-9999-0-s003.docx (592.6KB, docx)

Table S1: Clinical characteristics of participants providing plasma samples.

Table S2: Metabolite concentrations in plasma samples.

Table S3: Clinical characteristics of participants providing urine samples.

Table S4: Metabolite concentrations in urine samples.

Table S5: Clinical characteristics of participants providing fecal samples.

Table S6: Bacterial taxon abundances in fecal samples.

Table S7: qPCR primers.

CAS-9999-0-s002.zip (112.4KB, zip)

Data Availability Statement

Research data are not shared.


Articles from Cancer Science are provided here courtesy of Wiley

RESOURCES