Background & Summary

Cholelithiasis, the formation of calculi within the gallbladder or biliary tract, is a prevalent gastrointestinal disorder1 that affects 10–20% of the global adult population and poses a significant health care burden2. Most of the stones are cholesterol-based, resulting from bile compositional imbalances and impaired gallbladder motility, which promote cholesterol supersaturation, nucleation, and crystal growth2. Although frequently asymptomatic, cholelithiasis can lead to severe complications, such as biliary colic, acute cholecystitis, choledocholithiasis, and cholangitis, often requiring urgent intervention3. Laparoscopic cholecystectomy remains the definitive treatment for symptomatic disease. Despite extensive research, the intricate interplay among genetic predispositions, metabolic factors, diet, the gut microbiota, and gallbladder dysmotility in gallstone pathogenesis has not yet been fully elucidated4,5,6. Consequently, effective strategies for primary prevention and nonsurgical management remain limited7.

Metabolomics has emerged as a powerful tool for elucidating disease mechanisms and discovering biomarkers, offering direct insight into the functional phenotype of cholelithiasis through the capture of dysregulated metabolic pathways. Recent advances highlight the potential of metabolomics to elucidate the complex interplay between host metabolism and gallstone pathogenesis8. Within traditional Chinese medicine (TCM), cholelithiasis manifests as distinct syndrome patterns, notably liver Qi stagnation (LQS) and liver-gallbladder damp-heat (LGDH), each of which are hypothesized to represent a unique pathophysiological state. In Western medicine, the enrolled patients are classified as having acute calculous cholecystitis, whereas in TCM nosology, they are collectively referred to as having cholelithiasis (胆石症) and further differentiated into distinct syndrome patterns. However, the metabolic signatures that underpin these TCM syndromes remain largely unexplored.

To address this knowledge gap, we performed an integrated metabolomics analysis of serum and urine from cholelithiasis patients stratified according to TCM syndromes. The present analytical strategy combined untargeted profiling with three targeted liquid chromatography–tandem mass spectrometry (LC–MS/MS) platforms designed to quantify key metabolite classes, including amino acids, bile acids, fatty acids, and gut microbial metabolites.

Previous metabolomics studies of cholelithiasis have focused mainly on limited serum biomarkers and lack integration with TCM syndromic differentiation. The present dataset integrated untargeted and targeted LC–MS/MS profiling across serum and urine matrices in well-defined LQS and LGDH subgroups, enabling a systems-level view of metabolic heterogeneity. The present findings provide a comprehensively annotated metabolomic dataset from LQS and LGDH patients, featuring untargeted analysis of 602 serum and 636 urinary metabolites alongside targeted quantification of 210 serum and 204 urinary metabolites, augmented by detailed clinical metadata (Fig. 1). The present study provides LC–MS/MS data, sample processing methods, and quality control for the dataset. All the data are publicly available to facilitate hypothesis generation, biomarker validation, and cross-disciplinary research into TCM syndrome biology. Ultimately, the present study aimed to advance the molecular understanding of gallstone pathogenesis within TCM frameworks and support the development of personalized therapeutic interventions.

Fig. 1
Fig. 1
Full size image

Workflow of the cohort study and metabolomics analysis.

Methods

Inclusion and exclusion criteria for the cholelithiasis cohort

In total, 91 participants, comprising healthy controls (HC group, n = 33), patients with cholelithiasis with liver Qi stagnation (LQS group, n = 30), and patients with cholelithiasis with liver-gallbladder damp-heat (LGDH group, n = 28), were enrolled in the present study. The study protocol was approved by the Medical Ethics Committee of the First Affiliated Hospital of Dalian Medical University (PJ-KS-KY-2022-311). Participants were recruited from the inpatient and outpatient departments of the First Affiliated Hospital of Dalian Medical University through clinical assessment by attending physicians. All participants provided written informed consent for both their participation in the study and the publication of their anonymized data in public repositories. To protect participant privacy, all personal identifiers (such as names and ID numbers) were removed and replaced with unique study codes (e.g., SRX1, DZN1) prior to data analysis and sharing.

The present study enrolled patients with acute calculous cholecystitis in the context of gallstone disease. The diagnosis was established according to published clinical criteria9. Briefly, patients were considered to have acute calculous cholecystitis if they met the following conditions: (1) clinical symptoms, such as right upper abdominal pain, vomiting, jaundice, and fever; (2) a positive Murphy sign on physical examination, with possible local inflammatory findings, such as a palpable mass and tenderness in the right upper abdomen; and (3) supportive laboratory and imaging findings, including elevated C-reactive protein levels, increased white blood cell counts indicating systemic inflammation, and ultrasound, CT, or MRI evidence of gallstones and gallbladder inflammation. When either local inflammatory manifestations or systemic inflammatory manifestations were present, the likelihood of acute cholecystitis was considered high; when these clinical and laboratory findings were accompanied by positive imaging results, a diagnosis of acute cholecystitis was confirmed9.

Within the framework of traditional Chinese medicine (TCM), these patients were categorized under the broader disease entity of cholelithiasis. The diagnostic criteria for TCM syndrome patterns were obtained from the Guidelines for Consensus Opinion on the Integrated Diagnosis and Treatment of Cholelithiasis by Traditional Chinese and Western Medicine10.

Patients were classified as having the LQS pattern if they presented with one of the following combinations: (1) at least two major clinical symptoms plus at least one minor clinical symptom, together with the characteristic tongue and pulse presentation; or (2) at least one major clinical symptom plus at least three minor clinical symptoms, together with the characteristic tongue and pulse presentation. The major clinical symptoms included severe pain on the right side of the chest that may radiate to the shoulder and back and cause discomfort, loss of appetite, and pain that worsened with emotional agitation (e.g., anger). The minor clinical symptoms included chest tightness, belching (possibly accompanied by nausea), bitter taste in the mouth, dry throat, and incomplete defecation. The tongue and pulse presentation was characterized by a pale-red tongue with a thin white coating and a tight, wiry pulse.

Patients were classified as having the LGDH pattern if they presented with one of the following combinations: (1) at least two major clinical symptoms plus at least one minor clinical symptom, together with the characteristic tongue and pulse presentation; or (2) at least one major clinical symptom plus at least three minor clinical symptoms, together with the characteristic tongue and pulse presentation. The major clinical symptoms included right-sided chest or upper abdominal pain that was resistant to pressure and radiated to the right shoulder, yellowish urine, loose stools or constipation, chills, fever, and yellowing of the skin and eyes. The minor clinical symptoms included sore throat, dry and sticky mouth, abdominal distension, poor appetite, general fatigue, a sensation of heaviness, and nausea with an urge to vomit. The tongue and pulse presentation was characterized by a red tongue with a yellow, greasy coating and a tense, rapid pulse.

In summary, patients were considered eligible if they met the following criteria: (1) were over 18 years old; (2) fulfilled the diagnostic criteria of the TCM symptoms for CHOLE; and (3) provided written consent to participate in the study and able to undergo questionnaire surveys and cooperate with relevant laboratory tests. The exclusion criteria included (1) infectious diseases or severe trauma; (2) pregnant or lactating women; (3) severe heart, liver, or kidney dysfunction, thyroid diseases, severe anaemia, or malignant tumours; and (4) refusal to participate in the study or request the withdrawal of informed consent.

The study population was planned to consist of an approximately equal distribution of male and female subjects, and the individuals failing to meet the predefined eligibility criteria were systematically screened and excluded from the trial. For each enrolled subject, the general medical and biochemical examination parameters were recorded. The demographic parameters for the present cohort are summarized in Table 1 and Supplementary Table S1.

Table 1 Baseline information for the cholelithiasis cohort.
Full size table

Blood was collected in plastic vacutainer blood tubes, allowed to clot for 40 minutes, and then centrifuged at 13000 g for 10 minutes. The serum samples were frozen on dry ice immediately after collection and then stored at −80 °C until analysis.

A 0.5–1.0 mL sample of midstream urine was collected. After collection, the urine sample was not exposed to room temperature for more than 2 hours and was not stored at 4 °C for more than 4 hours. The samples were immediately aliquoted and frozen and stored at −80 °C.

Serum and urine sample pretreatment for metabolite analysis

Serum and urine samples were thawed at 4 °C and gently homogenized by pipetting. Afterwards, 120 μL of each biofluid was transferred to separate 2 mL microcentrifuge tubes (Aixgen, USA), followed by the addition of 480 μL of methanol. After vortexing for 5 min, the samples were incubated at room temperature for 10 min. Centrifugation was then performed at 13,000 × g for 10 min at 4 °C (Thermo Scientific, USA). The resulting supernatants were transferred to 1.5 mL tubes, lyophilized for 3 h (Labconco Corporation, USA), and stored at 4 °C until analysis. Prior to analysis, the dried extracts were reconstituted in 500 μL of acetonitrile-water (1:3, v/v). Quality control (QC) samples were prepared by pooling equal volumes of individual serum or urine samples.

Sample pretreatment for targeted metabolomics

The targeted metabolomics detection method was performed using the deep targeted metabolomics analysis platform established by our laboratory10.

Sample pretreatment for amino compounds

A mixed external standard stock solution containing authentic standards, including amino acids, nucleosides and amine derivatives, was serially diluted to generate 10-point mixed stock solution curves for quantitative analysis. For sample preparation, 50 μL of each standard was transferred to a 96-well plate, followed by the addition of 10 μL of HILIC-POS IS (M1-IS) and 190 μL of methanol. After vortexing (1,500 rpm, 3 min), the supernatants were centrifuged (5,300 rpm, 20 min) prior to LC–MS/MS analysis.

Sample pretreatment for bile acids and fatty acids

Samples (50 μL) were transferred to microcentrifuge tubes and mixed with 20 μL of internal standard (M2-IS) and 180 μL of methanol-acetonitrile (1:1, v/v). After the samples were vortexed (1,300 rpm, 5 min) and centrifuged (12,000 g, 20 min), 180 μL of the supernatant was transferred to a new vial for LC–MS/MS analysis. A mixed standard stock solution containing authentic bile acid and fatty acid standards was serially diluted to ten concentrations to generate 10-point external calibration curves for quantification.

Sample pretreatment for metabolites derived from the gut microbiota

Samples (50 μL) were transferred to a 750 μL 96-well plate, mixed with 10 μL of a gut metabolite internal standard (M4-IS) and 190 μL of methanol-acetonitrile (1:1, v/v). After the samples were vortexed (1,500 rpm, 3 min) and centrifuged (5,300 rpm, 20 min), 120 μL of the supernatant was diluted with 120 μL of ultrapure water. The mixture was revortexed (1,500 rpm, 3 min) and recentrifuged (5,300 rpm, 20 min), and 180 μL of the resulting supernatant was transferred to a new vial for LC–MS/MS analysis. A mixed standard stock solution containing authentic microbial metabolite standards was serially diluted to ten levels to generate 10-point external calibration curves for quantitative determination.

All targeted metabolite analyses shared a consistent quantification strategy based on 10-point external calibration curves and isotopically labelled internal standards for signal correction. Detailed information on the standard compositions, concentration ranges, and LC–MS/MS parameters is provided in Supplementary Table S2 (Tables S2–1 to S2-6).

Instrument parameters for untargeted metabolomics

Untargeted metabolomics profiling was conducted using an Ultimate 3000 UPLC system coupled to a Q Exactive quadrupole-Orbitrap mass spectrometer (Thermo Scientific, US). The separations employed two complementary columns.

For metabolite class 1, an Acquity™ HSS C18 column (2.1 × 100 mm; Waters) was used, with an injection volume of 5 μL. Mobile phase A consisted of 0.1% formic acid/water, and mobile phase B consisted of 0.1% formic acid/acetonitrile. The mobile phase gradient was as follows: 0–2 min (2% B) and 2–10 min (2–98% B).

For metabolite class 2, and Acquity™ BEH C18 column (1.7 μm, 2.1 × 100 mm; Waters) was used. Mobile phase A consisted of water, and mobile phase B consisted of acetonitrile/water (95:5, v/v). Both mobile phases contained 400 mg/L ammonium bicarbonate. The mobile phase gradient was as follows: 0–2 min (2% B), 2–10 min (2–100% B), followed by 5 min of washing/equilibration.

The shared chromatographic conditions consisted of 0.4 mL/min flow rate and 50 °C column temperature.

Mass spectrometric detection was performed using a heated electrospray ionization (HESI) source with the following parameters: a sheath gas flow rate of 45 arb, an auxiliary gas flow rate of 10 arb, a capillary temperature of 320 °C, and a vaporizer temperature of 355 °C. The S-Lens RF level was set to 55%. Full-scan MS data were acquired in positive/negative switching mode over an m/z range of 70–1000 with a resolution of 70,000 FWHM. For data-dependent MS/MS acquisition (dd-MS²), precursors were isolated using a 1.0 m/z window and fragmented via collision-induced dissociation (CID) with ultrapure nitrogen as the collision gas. MS² spectra were collected at 17,500 FWHM resolution. Dynamic exclusion (duration: 10 s), isotope exclusion, and TopN peak prioritization were enabled to increase metabolite identification.

Instrument parameters for targeted metabolomics

Targeted metabolite quantitation was performed using a Shimadzu LC-20ADXR system coupled to a Sciex 5500 + triple quadrupole mass spectrometer with TurboV™ electrospray ionization. Data acquisition employed scheduled multiple reaction monitoring (MRM). Three complementary chromatographic methods were utilized.

Amino compounds

Amino compounds were separated using an Acquity UPLC BEH Amide column (1.7 μm, 2.1 × 100 mm), with mobile phases A (90% aqueous acetonitrile) and B (50% aqueous acetonitrile. Both mobile phases were supplemented with ammonium formate/formic acid. The mobile phase gradient consisted of 0% → 70% B gradient over 15 min at a 0.3 mL/min flow rate (2 μL injection).

Bile acids and fatty acids

The bile acids and fatty acids were separated on a Hypersil GOLD column (1.9 μm, 2.1 × 100 mm) using mobile phases A (water + 2 mM ammonium acetate) and B (acetonitrile), with a multistep gradient (0.5–12 min: 0 → 17% B; 12–15.5 min: 17 → 30% B; 15.5–16.5 min: 55% B; 16.5–24.5 min: 55 → 95% B; and 24.5–27 min: 95% B) at 0.4 mL/min (3 μL injection).

Metabolites derived from gut microbiota

The metabolites derived from the gut microbiota were analysed on an ACE Excel PFP-C18 column (2 μm, 2.1 × 100 mm) using mobile phases A (0.1% formic acid/water) and B (acetonitrile), with a 5% → 95% B gradient over 9.5 min at 0.3 mL/min and 40 °C (5 μL injection). Common MS parameters included curtain gas (35 psi), ion gases 1/2 (50 psi), heater temperature (550 °C), and ion spray voltages (+5.5 kV [positive] and –4.5 kV [negative]).

Data extraction, processing and normalization

Metabolic peak alignment and extraction were conducted using Compound Discoverer software (Thermo Scientific). Structural annotation adhered to MSI11, leveraging the iPhenomeTM SMOL high-resolution MS/MS spectrum library12, alongside reference databases, including NIST 17 Tandem MS/MS13, mzCloud14, and the Human Metabolome Database (HMDB)15. Kyoto Encyclopaedia of Genes and Genomes16 (KEGG) was used exclusively for pathway mapping. The peak areas of the annotated metabolites were quantified using TraceFinder software (Thermo Scientific). Metabolite identification required a mass accuracy within ± 5 ppm. Chemical formulas were confirmed on the basis of exact mass, supported by isotopic pattern matching as follows: at least one isotope within 10 ppm mass error and a minimum 70% fit to the theoretical relative isotope abundance. Structural assignments were further validated using retention time and high-resolution MS/MS spectral similarity.

Targeted metabolites were acquired in selective reaction monitoring (SRM) mode and quantified using the isotope-labelled internal standard method. Primary and secondary mass spectrometry data for the targeted metabolites were compared against those of authentic standards. Quantitative data for each sample were calculated by integrating the standard curve with the corresponding area under the curve (AUC) values. Finally, calibration was performed using the internal standards. The analyses described above were executed utilizing Analyst® or MultiQuant™ software (version B.06.00; SCIEX, Singapore).

For untargeted metabolomics, QC-based LOESS normalization was performed to correct signal drift. For the targeted datasets, absolute concentrations were calculated from 10-point external calibration curves and normalized to the corresponding isotopically labelled internal standards.

Data Analysis

Metabolite feature-level quality control

To ensure data quality and reliability, the raw metabolomics data matrices from blood and urine samples were subjected to a series of stringent feature-level QC procedures.

Feature filtering based on missing values

Filtering based on Missingness in QC Samples: Given that QC samples are created by pooling all biological samples, they are theoretically expected to contain all stably detectable metabolites. Therefore, any metabolite feature with a missing value (NA) in any QC sample was immediately removed. This step was designed to rigorously eliminate features with unstable signals or those below the lower limit of detection of the instrument.

Filtering based on Within-Group Missingness: To preserve features stably expressed under specific biological conditions, we implemented a groupwise filtering strategy for the biological samples. A feature was removed if its proportion of missing values exceeded a predefined threshold (20%) within any single biological group. This approach aimed to eliminate potential confounding effects on statistical analysis arising from biologically specific absences rather than from technical error.

Feature selection by CV and Data imputation

Assessment and Filtering based on the Coefficient of Variation (CV): Following the missing value filtering, the coefficient of variation (CV) was calculated for all remaining features across the QC samples, serving as the primary metric for technical reproducibility. We established a CV threshold of 30%; any feature exceeding this threshold was discarded to ensure that the final dataset for analysis exhibited high technical stability.

Missing Value Imputation: To construct a complete data matrix for downstream analysis, the remaining random missing values within the biological samples were imputed using the K-Nearest Neighbours (KNN) algorithm. To prevent the introduction of artificial intergroup variance, the imputation process was performed independently within each biological group. To ensure the robustness of the workflow, a fallback mechanism was implemented; if KNN imputation failed for a specific group because of an insufficient sample size, the system automatically switched to groupwise median imputation.

Evaluation of QC efficacy

The effectiveness of the data-cleaning pipeline was systematically evaluated using several visualization methods.

Principal Component Analysis (PCA): PCA was performed on the final log10-transformed data matrix. The overall data quality was assessed by examining the tightness of the QC sample cluster on the score plot and by observing whether the biological samples displayed distribution patterns consistent with their group assignments.

Monitoring QC Sample Stability: The scores of all the QC samples on the first principal component (PC1) were extracted and plotted to monitor for signal drift across the analytical run. Minimal fluctuations in these scores indicate excellent stability of the analytical system.

Assessment of Technical Precision: The technical precision of the analytical platform was quantified by generating scatter plots of metabolite intensities between pairs of QC samples and calculating their Pearson’s correlation coefficient (r). A high correlation (typically r > 0.95) serves as strong evidence of a highly stable analytical platform.

Reproducibility of internal standards: The reproducibility of detection was further examined by evaluating the parameters of internal standards across all samples, which revealed a median RSD of 17.38% for retention time (RT), an ion intensity of 16.43%, and a maximum observed RSD of m/z 0.00496% for mass stability. The complete results are provided in Supplementary Table S3.

Data Records

The complete metabolomics datasets generated in this study are available at the Metabolomics Workbench17 under accession numbers ST004186 through ST004189. The data are organized into four specific compressed ZIP files corresponding to the study design: Cholelithiasis.zip (ST004186) contains the untargeted serum metabolomics data; Cholelithiasis targeted.zip (ST004187) contains the targeted serum metabolomics data; Cholelithiasis targeted (urine).zip (ST004188) contains the targeted urine metabolomics data; and Cholelithiasis(urine).zip (ST004189) contains the untargeted urine metabolomics data. Each compressed file includes the raw data tables, sample metadata, and processed data matrices. Within the processed data matrices, the first column lists the specific metabolite names (Metabolites), and the subsequent columns represent individual samples labelled with clinical group identifiers (e.g., SRX, QZX, and DZX) alongside the quality control samples (QC1–QC5).

Data Overview

The integrated dataset comprises metabolic profiles from a total of 91 participants, including 33 healthy controls, 30 patients with LQS syndrome, and 28 patients with LGDH syndrome. Through untargeted profiling, the study quantified 602 serum metabolites and 636 urinary metabolites. Complementing this, the targeted metabolomics analysis provided reliable absolute quantification for 210 serum metabolites and 204 urinary metabolites. These annotated datasets cover key metabolite classes such as amino acids, bile acids, fatty acids, and gut microbial metabolites across both serum and urine matrices.

Technical Validation

More than 82% of the analytes in both the untargeted and targeted metabolomics datasets had a relative standard deviation (RSD) of less than 25% in the QC samples (Fig. 2A). QC samples were employed to assess technical precision and reproducibility. Unsupervised principal component analysis (PCA) revealed the main sources of variation within dataset18 (Fig. 2B), with the QC samples clustering tightly, indicating high analytical stability. The temporal stability of the first principal component for the QC samples is illustrated in Fig. 2C. Spearman correlation analysis between the first and third QC injections demonstrated a strong correlation (Fig. 2D). To assess consistency between targeted and untargeted analyses, we performed a Pearson correlation analysis on the shared metabolites identified in both datasets, which revealed strong consistency between the methods (Fig. 3).

Fig. 2
Fig. 2
Full size image

Technical validation data for the QC samples in the methods. (A) Distribution graph of the relative variation coefficients for the QC sample analytes for both untargeted and targeted metabolomics of blood and urine. (B) Principal component analysis (PCA) of cohort samples for both untargeted and targeted metabolomics of blood and urine. (C) The stability of the first principal component of QC samples over time for both untargeted and targeted metabolomics of blood and urine. (D) Spearman correlation analysis between the first and third sets of QC data.

Fig. 3
Fig. 3
Full size image

Pearson correlation analysis of the metabolites represented by the targeted (Y axis) and untargeted (X axis) metabolomics platforms. (A) Pearson correlation analysis between metabolite concentrations obtained from untargeted metabolomics and targeted quantitative analysis in serum samples. (B) Pearson correlation analysis between metabolite concentrations obtained from untargeted metabolomics and targeted quantitative analysis in urine samples.