A 2024 Nature Communications study of 1,153 matched tissue and biofluid samples from 560 participants across three hospitals in southern China identified an NMR-detectable esophageal squamous cell carcinoma (ESCC) metabolome shared by tumor tissue, serum, and urine, and showed that any five of 16 serum or 10 urine metabolic signatures outperform clinical serological tumor markers for early-stage ESCC detection (AUC 0.984 and 0.930). Whether these simplified NMR biofluid panels are valid tools for population-level ESCC early detection remains unestablished: they have not been tested prospectively outside southern China, not in precursor-lesion and high-risk populations, not under rigorous multi-region statistics with control of confounding factors, and the cellular origins of the panel metabolites (tumor-intrinsic versus systemic or microbiome-derived) are unknown. The open problem is to establish the clinical validity and generalizability of these panels and to determine the biological origins of their constituent metabolites.

Background

Esophageal squamous cell carcinoma (ESCC), the dominant histological subtype of esophageal cancer worldwide, is usually diagnosed at an advanced stage and has poor survival; in high-incidence regions such as southern and northern China, endoscopic screening exists but is costly and underused, motivating non-invasive early-detection tests. Metabolomics simultaneously measures hundreds of small molecules (amino acids, lipids, organic acids) in tissues and biofluids; nuclear magnetic resonance (NMR) spectroscopy is a low-cost, highly reproducible metabolomics platform suited to high-throughput clinical screening, while mass spectrometry (MS) offers broader metabolite coverage. 'Liquid biopsy' here refers to detecting tumor-associated molecular signals in accessible fluids such as serum and urine rather than in tumor tissue itself. In the source study (Zhao et al., Nature Communications 15:2463, 2024), NMR- and MS-based metabolomics of 1,153 matched ESCC tissues, normal mucosae, and pre- and one-week post-operative sera and urines from 560 participants across three hospitals, combined with machine learning and WGCNA (weighted gene co-expression network analysis, a correlation-network method that clusters variables into co-varying modules — here applied to metabolites to link tissue and biofluid signatures), showed that aberrations in 'alanine, aspartate and glutamate metabolism' pervade ESCC evolution and are reflected in 16 serum and 10 urine metabolic signatures. Simplified NMR panels of any five serum or urine metabolites outperformed conventional serological tumor markers (areas under the receiver-operating curve, AUC, of 0.984 for serum and 0.930 for urine) and distinguished early-stage ESCC in a test set with accuracy 0.994 (serum) and 0.879 (urine); the study also noted that pre-diagnosis levels of several metabolites in the UK Biobank NMR ('Nightingale') platform were associated with future esophageal malignancy. The authors themselves state the limitations that gate clinical implementation: the cohort was drawn only from southern China; early-stage sample size was limited and the study relied on internal and external validation rather than systematic multiple-testing correction, so large-scale prospective cohorts in different regions under confounder control are needed; the panels must be validated in high-risk and precursor-lesion populations — low-grade and high-grade intraepithelial neoplasia (LGIN/HGIN, non-invasive stages of squamous dysplasia in the esophageal lining), gastroesophageal reflux disease (GERD), and Barrett's esophagus (metaplastic change of the esophageal lining) — with follow-up data to build a metabolic timeline; and integrated multi-omics (metabolome, proteome, microbiome) are needed to shed light on the metabolites' origins and functional mechanisms, work the authors describe as ongoing. Later literature provides adjacent but non-resolving progress: an independent two-cohort LC-MS and transcriptomic study including an LGIN follow-up cohort proposed a different nine-blood-metabolite biomarker set for early ESCC (Shen et al., 2026); an independent LC-MS study in northern China reported a seven-metabolite fatty-acid panel with AUC 0.804 (Li et al., 2026); prospective UK Biobank NMR analyses found metabolite-based risk scores predicting gastrointestinal cancer incidence with only moderate accuracy (Sun et al., 2026; You et al., 2025); Mendelian randomization (a genetic-instrument method using germline variants as proxies for exposures to test causality) found limited causal metabolite-esophageal-disease relationships (Yang et al., 2024); and longitudinal plasma metabolomics has been applied to ESCC treatment monitoring rather than early detection (Qian et al., 2026). None of these prospectively validates the simplified NMR five-metabolite panels across populations or traces the cellular origins of their metabolites.

Problem Statement

Establish whether the NMR-based simplified biofluid metabolite panels — any five of the 16 serum or 10 urine metabolic signatures identified by Zhao et al. (Nature Communications 15:2463, 2024) — are valid, generalizable tools for early detection of esophageal squamous cell carcinoma at the population level, and determine the biological origins of their constituent metabolites. In keeping with the source study's stated agenda, the problem decomposes into three parts: (1) whether the panels are valid and generalizable across large-scale populations and regions beyond southern China, with systematic control of confounders and rigorous multiple-testing correction across all discovery and validation steps; (2) the performance of the panels in high-risk and precursor-lesion populations — low-grade and high-grade intraepithelial neoplasia, GERD, and Barrett's esophagus — including longitudinal follow-up establishing the panels' predictive value for future ESCC onset and their behavior along the progression timeline; and (3) the origins and functional mechanisms of the panel metabolites, and the extent to which the biofluid signals are tumor-cell intrinsic, derived from the tumor microenvironment or systemic host response, or microbiome-associated. These three parts jointly serve one overarching objective — determining whether the panels are biologically interpretable and clinically robust enough for population-level ESCC early detection — and they are jointly posed: independently establishing one of them, such as a mechanistic attribution without evidence on the panels' validity and generalizability or a single-region replication without the others, does not resolve the objective. Because the relevant prospective multi-region cohorts, longitudinal high-risk follow-up, and integrated multi-omics datasets do not yet exist, the question turns on weighing the evidence already available about the panels' validity and the proposed origins rather than on the collection of new data.

References

  1. Yan Zhao, Changchun Ma, Rongzhi Cai, Lijing Xin, Yongsheng Li, Lixin Ke, Wei Ye, Ting Ouyang, Jiahao Liang, Renhua Wu, Yan Lin. NMR and MS reveal characteristic metabolome atlas and optimize esophageal squamous cell carcinoma early detection. Nature Communications 15:2463 (2024). DOI: 10.1038/s41467-024-46837-0. https://doi.org/10.1038/s41467-024-46837-0
  2. Xia Shen, Yaqi Zhang, Qinqin Wang, Xumiao Li, Min Gao, Jun Li, Chen Li, Hui Wang. Comprehensive metabolic profiling uncovers early variation and biomarkers of esophageal squamous cell carcinoma. Science China Life Sciences 69(3):881-895 (2026). DOI: 10.1007/s11427-025-3082-9. https://doi.org/10.1007/s11427-025-3082-9
  3. Yonghao Li, Lu Zhou, Ruixue Huang, Zhongbing Wu, Jiachang Guan, Xuelei Qiao, Yiran Kong, Yihang Gan, Yushuang Zhang, Jing Li. Serum Untargeted Metabolomics Profiling of Esophageal Squamous Cell Carcinoma in High-Incidence Areas of China. International Journal of General Medicine 19:608099 (2026). DOI: 10.2147/IJGM.S608099. https://doi.org/10.2147/IJGM.S608099
  4. Ming Sun, Weiwei Chen, Xueni Cheng, Hengyu Cui, Jianv Huang, Rong Huang, Yumeng Li, Yanan Ma, Dong Hang. Comprehensive Evaluation of Lipids, Lipoproteins, and Metabolites Associated with Gastrointestinal Cancer Risk. Cancer Epidemiology, Biomarkers & Prevention 35(2):225-236 (2026). DOI: 10.1158/1055-9965.EPI-25-1301. https://doi.org/10.1158/1055-9965.EPI-25-1301
  5. Hanlei Yang, Yulan Wang, Yuewei Zhao, Leiqun Cao, Changqiang Chen, Wenjun Yu. Causal effects of genetically determined metabolites and metabolite ratios on esophageal diseases: a two-sample Mendelian randomization study. BMC Gastroenterology 24:310 (2024). DOI: 10.1186/s12876-024-03411-8. https://doi.org/10.1186/s12876-024-03411-8
  6. Jia You, Xi-Han Cui, Yi-Lin Chen, Yi-Xuan Wang, Hai-Yun Li, Yi-Xuan Qiang, Ji-Yun Cheng, Yue-Ting Deng, Yu Guo, Peng Ren, Yi Zhang, Yu He, Xiao-Yu He, Shi-Dong Chen, Ya-Ru Zhang, Yu-Yuan Huang, Ying Mao, Jian-Feng Feng, Wei Cheng, Jin-Tai Yu. Mapping the plasma metabolome to human health and disease in 274,241 adults. Nature Metabolism 7(11):2366-2384 (2025). DOI: 10.1038/s42255-025-01371-1. https://doi.org/10.1038/s42255-025-01371-1
  7. Zheng-Yu Qian, Yuqing Deng, Mian Xi, Xin-Yao Zhang, Hui Sun, Minjie Mao, Xiao-Jing Luo, Mengmeng Li, Qihua Zhang, BaoQing Chen, Yi-Qian Pan, Jia Liu, Ming-Ming He, Hao-Jie Chen, Chiu-Wing Yeung, Yashang Zheng, Jia-Qian Huang, Meng Yang, Zong-Jian Li, Zhao-Lei Zeng, Huai-Qiang Ju, Rui-Hua Xu, Huiyan Luo. Longitudinal Plasma Metabolomics Guides Dynamic Risk Assessment and Dietary Modulation for Esophageal Squamous Cell Cancer Chemoimmunotherapy. Cancer Discovery 16(7):1341-1359 (2026). DOI: 10.1158/2159-8290.CD-25-1449. https://doi.org/10.1158/2159-8290.CD-25-1449
  8. Hexin Ma, Xinran Zhao, Zhihui Li, Zhen Zhang, Xinhua Pan, Mengxuan Du, Qingqing Wan, Chaojun Zhang, Xu Wang, Hongyu Zhao, Zhien Feng, Xin Zhang, Wan-Tao Chen. Serum metabolomics profiling in head and neck squamous cell carcinoma: a multicentre diagnostic study. EBioMedicine 130:106379 (2026). DOI: 10.1016/j.ebiom.2026.106379. https://doi.org/10.1016/j.ebiom.2026.106379