NMR-based metabolomics of a single blood sample can separate patients with solid tumors from those without cancer in mixed populations referred with nonspecific symptoms (Larkin et al., 2022, AUC 0.83 in an independent test set, but only 24 cancers, one centre). The open translational problem is to establish whether this signal can become a deployable diagnostic test: validate it in substantially larger, multi-site prospective cohorts with per-cancer-type sensitivity and clinically tuned thresholds; demonstrate (or refute) clinical utility and cost-effectiveness as a triage tool; determine the mechanistic origin of the circulating metabolite changes (direct tumor metabolism versus indirect systemic host response); and establish how the test should be combined with or compared against circulating tumor DNA (ctDNA)-based assays. A 2026 prospective single-centre follow-up (SCAN2) validated the signature in a more comorbid cohort and added glycomics, but the large-scale validation, utility, mechanism, and ctDNA-integration questions all remain open.

Background

Early cancer diagnosis improves outcomes, but many cancers announce themselves only through nonspecific symptoms such as weight loss, fatigue, or unexplained pain — complaints that fit no single organ-specific referral route. In England, so-called multidisciplinary diagnostic centres (MDCs) were created for this 'low-risk, but not no-risk' group: patients in whom cancer is plausible but too unlikely to justify urgent site-specific investigation. The Oxfordshire Suspected CANcer (SCAN) pathway is one such MDC (Nicholson et al., 2018, BMJ Open). Even in MDCs, however, the initial work-up rests largely on imaging, so a cheap, rapid blood test that helps triage these patients would be valuable.

One candidate technology is proton nuclear magnetic resonance (NMR) metabolomics: benchtop-compatible spectroscopy of serum that in one measurement quantifies dozens of small molecules (amino acids, glucose, lactate) plus a lipoprotein profile. The Oxford group had previously shown in mice that biofluid metabolomics detects very small tumors and grades metastatic burden, with tumor-site-dependent systemic profiles that already suggested the signal is not purely tumor-intrinsic (Larkin et al., 2016, Theranostics). In the study at issue (Larkin et al., 2022, Clinical Cancer Research), serum from 304 SCAN patients was analyzed by NMR, and orthogonal partial least squares discriminant analysis (OPLS-DA — a supervised multivariate projection method that separates predefined patient classes) classified patients by their eventual MDC diagnosis. On an independent test set, the model identified solid tumors with area under the ROC curve (AUC, a threshold-free summary of sensitivity versus specificity) of 0.83 (maximum sensitivity 94%, specificity 82%), and within the cancer patients it flagged metastatic disease with 94%/88% sensitivity/specificity. In two patients the baseline metabolomics result preceded a CT-based cancer diagnosis.

The authors themselves identified four unresolved issues. First, scale: only 24 solid tumors (8.2% of 294 analyzed patients) of highly mixed types were included, hematologic malignancies were excluded, and per-cancer-type sensitivity, cross-site robustness, and operating thresholds for specific clinical uses are unknown. Second, utility: whether adding metabolomics to the pathway reduces time-to-diagnosis, unnecessary imaging, cost, and radiation exposure — and the practical logistics of centralized NMR analysis — has never been measured. Third, mechanism: the discussion argues an 'indirect magnification' hypothesis, in which the circulating changes arise substantially from the host response (liver and adipose metabolism, inflammatory cytokines such as interleukin-6 (IL6) and interleukin-1β (IL1β), reactive stroma) rather than from the tumor's own metabolism; small tumors are too small to perturb whole-body metabolite pools directly, and animal work supports site-dependent systemic effects (Larkin et al., 2016; isotope-tracing in zebrafish shows alanine shuttling between melanoma and liver, Naser et al., 2021). This mechanistic question is unresolved in humans. Fourth, integration: ctDNA assays (sequencing of DNA shed by tumors into blood) and multi-cancer early detection (MCED) tests built on it are advancing in the same symptomatic population — the SYMPLIFY study evaluated the Galleri MCED blood test in over 6,000 patients referred with signs or symptoms of cancer in England and Wales (Nicholson et al., 2023, Lancet Oncology) — and an accompanying commentary framed metabolomics as a potential complementary, cheaper technology (Van Tine and Lyssiotis, 2022), but no head-to-head or combined evaluation exists. At the population scale, NMR metabolites measured in 83,290 UK Biobank participants are associated with subsequent cancer risk (Hu et al., 2024), supporting the biological signal's scalability without settling diagnostic performance in referred patients. The most direct follow-up, SCAN2 (Kacerova et al., 2026), extended the approach prospectively in the same single centre.

Problem Statement

Establish whether NMR-based blood metabolomics can become a clinically effective cancer-detection and triage test for patients referred with nonspecific symptoms, the open question left by Larkin et al. (2022) after their single-centre study of 294 analyzable patients (24 solid tumors). This single translational objective comprises four scientifically linked open questions: (1) Prospective validation at scale — test the metabolomics classifier in substantially larger, more diverse prospective cohorts, both in 'low-risk, but not no-risk' nonspecific-symptom referral pathways and in organ-specific referral streams, to determine overall and per-cancer-type sensitivity and specificity, to assess performance across sites and populations, and to refine operating thresholds appropriate to different clinical use cases. (2) Clinical utility — determine, in formal prospective studies, whether adding metabolomics to the diagnostic pathway reduces time-to-diagnosis, decreases unnecessary imaging with its associated costs and radiation exposure, and improves patient outcomes when used as a triage tool, together with cost-effectiveness and implementation logistics such as centralized NMR turnaround times and external laboratory integration. (3) Mechanism — determine to what extent the observed circulating metabolite changes arise from direct tumor metabolic activity versus indirect systemic host responses (for example altered liver metabolism or tumor-associated inflammatory cytokine signaling), via targeted experimental and translational studies. (4) Integration — determine how metabolomics should be combined with or compared against emerging modalities such as circulating tumor DNA assays, and how false positives should be managed given low disease prevalence. Negative results on any component (for example failure of performance to generalize across sites, absence of measurable utility benefit, or demonstration that the signal is dominated by a nonspecific host response) resolve that component as fully as positive ones. Each component is resolved by a positive or a negative finding.

References

  1. Larkin JR, Anthony S, Johanssen VA, Yeo T, Sealey M, Yates AG, Friedemann Smith C, Claridge TDW, Nicholson BD, Moreland J-A, Gleeson F, Sibson NR, Anthony DC, Probert F. 'Metabolomic Biomarkers in Blood Samples Identify Cancers in a Mixed Population of Patients with Nonspecific Symptoms.' Clinical Cancer Research 28(8):1651-1661 (2022). DOI: 10.1158/1078-0432.CCR-21-2855. https://doi.org/10.1158/1078-0432.CCR-21-2855
  2. Kacerova T, Yates AG, Larkin JR, Shulgin B, Miller JJJJ, Harris Gleave PL, de Jel S, Cheeseman J, Elgood-Hunt G, Al-Sattar H, Appleby N, Bao G, Batth G, Benamore R, Blake A, Ciuban G, Collins KK, de Vere Hunt I, Dreau H, Evans M, Gibb N, Gleeson F, Hayles S, Heath L, Jackson A, Johnson C, Kaveney Z, Lewis D, Lomas R, Mitchener D, Montagu A, Moreland J-A, Moore N, Nicholson B, Roberts S, Sadhoo N, Saujani S, Schuh A, Slater A, Smith S, Taylor N, Tonner S, Waterfield S, Wing L, Schiffer E, Spencer DIR, Anthony S, Anthony DC. 'NMR metabolomics and glycomics for cancer detection in patients with non-specific symptoms: a prospective observational cohort study.' eBioMedicine 131:106434 (2026). DOI: 10.1016/j.ebiom.2026.106434. https://doi.org/10.1016/j.ebiom.2026.106434
  3. Hu C, Fan Y, Lin Z, Xie X, Huang S, Hu Z. 'Metabolomic landscape of overall and common cancers in the UK Biobank: A prospective cohort study.' International Journal of Cancer 155(1):27-39 (2024). DOI: 10.1002/ijc.34884. https://doi.org/10.1002/ijc.34884
  4. Nicholson BD, Oke J, Virdee PS, Harris DA, O'Doherty C, Park JES, Hamady Z, Sehgal V, Millar A, Medley L, Tonner S, Vargova M, Engonidou L, Riahi K, Luan Y, Hiom S, Kumar H, Nandani H, Kurtzman KN, Yu L-M, Freestone C, Pearson S, Hobbs FDR, Perera R, Middleton MR. 'Multi-cancer early detection test in symptomatic patients referred for cancer investigation in England and Wales (SYMPLIFY): a large-scale, observational cohort study.' The Lancet Oncology 24(7):733-743 (2023). DOI: 10.1016/S1470-2045(23)00277-2. https://doi.org/10.1016/S1470-2045(23)00277-2
  5. Van Tine BA, Lyssiotis CA. 'In an Era of ctDNA, Is Metabolomics the New Kid on the Block?' Clinical Cancer Research 28(8):1477-1478 (2022). DOI: 10.1158/1078-0432.CCR-21-4180. https://doi.org/10.1158/1078-0432.CCR-21-4180
  6. Nicholson BD, Oke J, Friedemann Smith C, Phillips J-A, Lee J, Abel L, Kelly S, Gould I, Mackay T, Kaveney Z, Anthony S, Hayles S, Lasserson D, Gleeson F. 'The Suspected CANcer (SCAN) pathway: protocol for evaluating a new standard of care for patients with non-specific symptoms of cancer.' BMJ Open 8(1):e018168 (2018). DOI: 10.1136/bmjopen-2017-018168. https://doi.org/10.1136/bmjopen-2017-018168
  7. Larkin JR, Dickens AM, Claridge TDW, Bristow C, Andreou K, Anthony DC, Sibson NR. 'Early Diagnosis of Brain Metastases Using a Biofluids-Metabolomics Approach in Mice.' Theranostics 6(12):2161-2169 (2016). DOI: 10.7150/thno.16538. https://doi.org/10.7150/thno.16538
  8. Naser FJ, Jackstadt MM, Fowle-Grider R, Spalding JL, Cho K, Stancliffe E, Doonan SR, Kramer ET, Yao L, Krasnick B, Ding L, Fields RC, Kaufman CK, Shriver LP, Johnson SL, Patti GJ. 'Isotope tracing in adult zebrafish reveals alanine cycling between melanoma and liver.' Cell Metabolism 33(7):1493-1504 (2021). DOI: 10.1016/j.cmet.2021.04.014. https://doi.org/10.1016/j.cmet.2021.04.014