BIDMC Metabolomic Cohort of Adults with Biopsy-Proven MASLD

@article{PhysioNet-bidmc-metabolomic-masld-1.0.0,
  author = {{Gutierrez de Piñeres}, Valeria and Tamayo-Torres, Claudia S and Mantzoros, Christos},
  title = {{BIDMC Metabolomic Cohort of Adults with Biopsy-Proven MASLD}},
  journal = {{PhysioNet}},
  year = {2026},
  month = aug,
  note = {Version 1.0.0},
  doi = {10.13026/a91p-k288},
  url = {https://doi.org/10.13026/a91p-k288}
}

Abstract

Metabolic dysfunction-associated steatotic liver disease (MASLD), including its progressive form metabolic dysfunction-associated steatohepatitis (MASH), is the leading cause of chronic liver disease. Accurate delineation of the metabolites and signaling pathways that drive hepatic steatosis, inflammation and fibrosis remains essential to support further development of effective screening and treatment strategies.

This project presents a fully de-identified, high-dimensional serum metabolomics dataset derived from a prospective cross-sectional Gastroenterology–Hepatology cohort. Metabolomic analyses and data processing were conducted at Beth Israel Deaconess Medical Center (BIDMC), a tertiary academic center affiliated with Harvard Medical School. Fasted serum samples were analyzed using ultra-high-performance liquid chromatography coupled with tandem mass spectrometry (UPLC-MS/MS), generating 1,618 targeted metabolomic features per participant.

The dataset comprises 110 adult subjects presenting with abnormal liver function tests at a Gastroenterology–Hepatology clinic in the Sydney West Local Health District. This resource is designed to help validate findings from other studies and support hypothesis generation about metabolites involved in MASLD, as well as latent metabolic subgroups across the disease spectrum. The dataset is provided in CSV format with an accompanying README to facilitate secure, transparent, and reproducible reuse under appropriate research agreements.

Background

This database was created to support transparent, reproducible, and collaborative research on biopsy-proven metabolic dysfunction-associated steatotic liver disease (MASLD) using high-dimensional targeted metabolomics derived from fasted serum samples. MASLD represents one of the most prevalent chronic liver diseases worldwide and is a major contributor to liver-related morbidity and mortality [1]. Disease progression is driven by hepatic inflammation and fibrosis, which are the principal determinants of long-term outcomes [2].

Liver biopsy remains the reference standard for diagnosing MASH and staging fibrosis; however, its invasiveness, cost, and susceptibility to sampling variability limit its widespread applicability [3,4]. These limitations have driven increasing interest in non-invasive approaches for disease detection and stratification.

Metabolomics has emerged as a promising strategy for capturing systemic metabolic alterations associated with MASLD [5,6], enabling the identification of candidate biomarkers for metabolic clustering. In this context, high-dimensional mass spectrometry-based platforms offer the ability to profile a wide range of metabolites with high sensitivity and specificity.

By sharing a tidy, de-identified, mass-spectrometry-based omics resource, we aim to maximize its reuse potential for hypothesis generation and external validation.

Methods

Adults presenting with abnormal liver function tests at a Gastroenterology-Hepatology clinic in the Sydney West Local Health District underwent same-morning liver biopsy and fasted serum sample collection. Serum samples were profiled by Metabolon Inc. using UPLC-MS/MS to quantify 1,618 targeted metabolites, metabolic intermediates, lipids, amino-acid derivatives, organic acids, and small-molecule biomarkers. The resulting data were cleaned, exported into a tidy tabular structure, and fully de-identified under HIPAA Safe Harbor standards. All shared variables have complete coverage per subject, and no imputation was required.

Data Description

The dataset is provided as a single comma-separated values file, database.csv, containing 110 observations (rows excluding the header) and 1,619 columns. The file is UTF-8 encoded. The first column (ID) represents a unique identifier for each sample, while the remaining 1,618 columns correspond to metabolomic measurements obtained from fasting serum samples using a mass spectrometry–based targeted metabolomics platform.

Each metabolite column reflects the measured abundance of a specific small molecule, encompassing a wide range of biochemical classes including lipids, amino acids and their derivatives, organic acids, metabolic intermediates, vitamins, and other circulating biomarkers. When available, metabolite names follow standard biochemical nomenclature; however, a subset of variables is labeled using platform-specific identifiers (e.g., X-... entries), which correspond to detected but not fully characterized compounds, features with unresolved structural identity, or internal identifiers assigned by the analytical platform. These variables should be interpreted as distinct quantitative metabolite features despite lacking full annotation.

In addition, certain metabolite names include symbols such as “*” or “**”, which typically reflect platform-specific annotation conventions or confidence levels. The dataset does not include explicit measurement units for individual metabolites; therefore, values likely represent relative abundances or normalized signal intensities derived from mass spectrometry and are most appropriate for comparison across samples within each metabolite rather than as absolute concentrations.

Importantly, the dataset contains only metabolomic variables and does not include clinical or phenotypic labels. Specifically, no columns corresponding to biopsy-proven fibrosis stage, MASH status, or case/control classification are present in the file. Any terms within metabolite names that may appear clinically suggestive (e.g., fibrinopeptide-related compounds) refer strictly to biochemical entities and should not be interpreted as clinical outcome variables. Consequently, any linkage between metabolomic profiles and clinical phenotypes would require external metadata not included in this dataset.

Usage Notes

Aims and motivations for sharing this resource:

1. To enable hypothesis generation regarding metabolites and signaling pathways in biopsy-proven MASLD.
2. To support creation of classification algorithms that can translate mass-spectrometry-derived metabolomic features into a non-invasive, biopsy-free diagnostic screening pathway for future patient or participant identification and allocation.
3. To maximize reuse potential for cross-disciplinary collaboration and transparent research, reducing risk, cost, and complexity of biopsy-based screening for future non-invasive metabolic liver disease diagnostics.

This dataset is suitable for exploratory and hypothesis-generating analyses of serum metabolomic profiles in individuals with MASLD. It may be used for applications such as biomarker discovery, pathway analysis, dimensionality reduction, clustering, and the development of predictive models based on metabolomic features.

Several limitations should be considered when interpreting results. A portion of the underlying cohort was collected retrospectively; however, liver biopsies and serum samples were obtained independently of the present study aims, reducing the likelihood of analysis-related bias. Participants were recruited from secondary and specialized care settings, including hepatology and metabolic clinics, which represent higher-risk populations. This may limit the generalizability of findings to broader or community-based populations and should be taken into account when interpreting model performance or biological associations. At the same time, the dataset reflects real-world conditions in specialized clinical environments, which may be valuable for studies focused on high-risk groups.

The sample size is modest (n = 110), and the dataset is high-dimensional (1,618 metabolite variables), which increases the risk of overfitting in statistical or machine learning analyses. Robust validation strategies, including cross-validation or external validation in independent cohorts, are strongly recommended. In addition, although the underlying study includes biopsy-proven fibrosis staging and clinical phenotyping, these variables are not included in the shared dataset, which limits its use for supervised learning unless external metadata are available.

Metabolite measurements are provided without explicit units and likely represent normalized or relative abundances derived from mass spectrometry. Therefore, comparisons should be made across samples within each metabolite rather than across different metabolites. Finally, a subset of variables corresponds to unidentified or partially annotated metabolites (e.g., X-... features), which may limit direct biological interpretability but can still provide useful signals for computational analyses.

Overall, this dataset is best suited for exploratory analyses and methodological development, and findings should be interpreted in light of cohort composition and validated in larger, independent, and more diverse populations.

Release Notes

Version 1.0.0 – Initial release of the dataset, including fully de-identified multi-omics features.

Ethics

The study protocol for this cohort was reviewed and approved by institutional ethics committees in Australia and the United States. All adult participants, including healthy controls, provided written informed consent prior to serum collection and metabolomic profiling. Liver biopsies were performed on the morning of sample acquisition to confirm the diagnosis of metabolic dysfunction-associated steatotic liver disease (MASLD).

All shared files were fully de-identified in accordance with HIPAA Safe Harbor standards, contain no protected health identifiers, and are distributed exclusively for credentialed, agreement-based research reuse under appropriate data transfer and use agreements.

The non-human-subjects research determination applies to the dataset shared on PhysioNet, as the dataset is fully de-identified and does not contain any information that could be used to identify individual participants.

Exempt approval for this project was granted by the Director of Operations for the Committee on Clinical Investigations (CCI) at BIDMC (IRB protocol number: 2018D000086).

Conflicts of Interest

CSM reports grants through his institution from Amgen, Abbott, Merck, Massachusetts Life Sciences Center, and Boehringer Ingelheim, has been a shareholder of Coherus Inc; he reports personal consulting fees and support with research reagents from Ansh Inc., collaborative research support from LabCorp Inc., reports personal consulting fees from Genfit, Lumos, Amgen, Corcept, Aligos, Intercept, 89 Bio, Madrigal, Boehringer Ingelheim, Azurity, and Regeneron, reports travel support and fees from TMIOA, Elsevier, and the Cardio Metabolic Health Conference. None is related to the work presented herein. The remaining authors declare no competing interests.

References

1. Miller DM, McCauley KF, Dunham-Snary KJ. Metabolic dysfunction-associated steatotic liver disease (MASLD): mechanisms, clinical implications and therapeutic advances. Endocrinol Diabetes Metab. 2025;8(6):e70132. doi:10.1002/edm2.70132. PMID:41255342; PMCID:PMC12627968.
2. Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, Harrison SA, Brunt EM, Sanyal AJ. The diagnosis and management of nonalcoholic fatty liver disease: practice guidance from the American Association for the Study of Liver Diseases. Hepatology. 2018;67(1):328-357. doi:10.1002/hep.29367. PMID:28714183.
3. Kouvari M, Valenzuela-Vallejo L, Guatibonza-Garcia V, Polyzos SA, Deng Y, Kokkorakis M, Agraz M, Mylonakis SC, Katsarou A, Verrastro O, Markakis G, Eslam M, Papatheodoridis G, George J, Mingrone G, Mantzoros CS. Liver biopsy-based validation, confirmation and comparison of the diagnostic performance of established and novel non-invasive steatotic liver disease indexes: results from a large multi-center study. Metabolism. 2023;147:155666. doi:10.1016/j.metabol.2023.155666. PMID:37527759.
4. Stefanakis K, Mingrone G, George J, Mantzoros CS. Novel machine-learning models outperform noninvasive tests in metabolic dysfunction-associated steatohepatitis with fibrosis. Clin Gastroenterol Hepatol. 2025 Nov 7:S1542-3565(25)00938-3. doi:10.1016/j.cgh.2025.10.029. Epub ahead of print. PMID:41207519.
5. Huang Y, Li J, Song S, Du B, Cao Y, Wang Y, et al. Blood metabolic panels for identifying significant fibrosis and inflammation in patients with MASLD. Cell Rep Med. 2026 Jan 20;7(1):102522. doi:10.1016/j.xcrm.2025.102522. PMID:41421352; PMCID:PMC12866143.