Abstract

   Metabolic changes precede malignant histology. However, it remains
   unclear whether detectable characteristic metabolome exists in
   esophageal squamous cell carcinoma (ESCC) tissues and biofluids for
   early diagnosis. Here, we conduct NMR- and MS-based metabolomics on
   1,153 matched ESCC tissues, normal mucosae, pre- and one-week
   post-operative sera and urines from 560 participants across three
   hospitals, with machine learning and WGCNA. Aberrations in ‘alanine,
   aspartate and glutamate metabolism’ proved to be prevalent throughout
   the ESCC evolution, consistently identified by NMR and MS, and
   reflected in 16 serum and 10 urine metabolic signatures in both
   discovery and validation sets. NMR-based simplified panels of any five
   serum or urine metabolites outperform clinical serological tumor
   markers (AUC = 0.984 and 0.930, respectively), and are effective in
   distinguishing early-stage ESCC in test set (serum accuracy = 0.994,
   urine accuracy = 0.879). Collectively, NMR-based biofluid screening can
   reveal characteristic metabolic events of ESCC and be feasible for
   early detection (ChiCTR2300073613).

   Subject terms: Oesophageal cancer, Metabolomics, Diagnostic markers,
   Cancer screening
     __________________________________________________________________

   Metabolic changes often occur during the early stages of cancer
   development. Here, the authors develop metabolomics signatures from
   tissues, pre- and post-operative sera and urines in esophageal squamous
   cell carcinoma, which may aid in early diagnosis.

Introduction

   Esophageal cancer (EC) is a significant public health concern
   worldwide^[44]1. China accounts for over half of the global annual
   incidence and mortality, of which more than 90% are esophageal squamous
   cell carcinoma (ESCC), and the T0, Tis, and T1 stages account for 10.8%
   of the total cases^[45]2,[46]3. The cure rate for early-stage ESCC
   exceeds 90%. However, due to the subtle nature of symptoms and the lack
   of biomarkers for early diagnosis, most patients are typically
   diagnosed in late-stage T3-T4, resulting in a 5-year survival rate of
   only approximately 21%^[47]4. Currently, the gold standard for
   diagnosing ESCC primarily relies on endoscopy coupled with
   histopathology, but its invasive nature reduces patient
   compliance^[48]5. Barium swallow and CT imaging techniques involve
   radiation exposure and tend to miss small lesions. Conventional
   serological tumor markers, such as Cytokeratin-19-fragment CYFRA21-1
   (Crfr211), Squamous Cell Carcinoma Antigen (SCC), Carcinoembryonic
   Antigen (CEA), Carbohydrate Antigen 19-9 (CA 19-9) and Carbohydrate
   Antigen 15-3 (CA 15-3) have shown suboptimal accuracy in clinical
   practice. With the advancements in omics technologies, researchers have
   explored several novel biomarkers for ESCC diagnosis, including DNA
   methylation markers, serum miRNA, autoantibodies, somatic gene
   mutations, salivary exosomes, and artificial intelligence (AI)-assisted
   sponge cytology^[49]6–[50]11. However, these approaches face
   limitations due to the requirement of advanced technology
   platforms, methodological instability, or high costs, hindering their
   translation into large-scale clinical applications. Consequently, there
   is an urgent need to develop reliable, non-invasive, accessible and
   affordable tools for ESCC early detection^[51]12.

   It takes years for ESCC to progress from squamous cell hyperplasia to
   atypical hyperplasia, carcinoma in situ, early-stage and invasive
   cancer^[52]13. In addition, metabolic phenotypic changes could precede
   malignant histological alterations, providing a significant opportunity
   for early detection and timely intervention^[53]14. However, it remains
   to be established whether there are characteristic metabolic changes
   during ESCC evolution and whether such metabolic changes can be
   detected. Metabolomics holds great promise for identifying
   disease-associated metabolites, highlighting its valuable insights into
   early diagnosis, treatment strategies, and mechanistic
   investigations^[54]15. Proton nuclear magnetic resonance (^1H-NMR) and
   mass spectrometry (MS) are the most mainstream technological platforms
   in metabolomics^[55]16,[56]17. MS exhibits high sensitivity (SE) but
   requires expensive standard reagents. ^1H-NMR has shown remarkable
   stability, excellent reproducibility, quantitative nature, non-invasive
   sample analysis, and has been well-suited for clinical multi-center,
   large-scale, and longitudinal monitoring studies for establishing
   successful clinical applications in countries such as the United
   Kingdom and Canada^[57]18,[58]19.

   Most existing metabolomics studies of ESCC have mainly relied on
   analyzing biofluid samples, such as serum and urine. Our previous study
   showed that NMR-based biofluid metabolomic profiles can discriminate
   ESCC patients from healthy controls (HCs), suggesting the potential
   utility of biofluid metabolic fingerprinting as predictors for
   ESCC^[59]20–[60]22. However, potential confounders, such as
   environmental factors, lifestyle habits, phenotypic variations and
   comorbidities, might influence biological fluid metabolism, leading to
   a gap between ESCC biofluids and characteristic molecular events of
   ESCC tissues. In this study, we employed a comprehensive research
   strategy incorporating 1,153 multi-dimensional matched specimens, NMR
   and targeted MS cross-platform testing, as well as multi-center
   validation. We aimed to investigate tumor tissue-specific metabolic
   biomarkers during ESCC evolution, and then leverage them as references