Metabolomics: Plasma Markers Track Lung Cancer Immunotherapy
PD-L1 staining still drives most checkpoint inhibitor decisions in advanced non-small cell lung cancer (NSCLC), yet some PD-L1-high patients progress early while some PD-L1-negative patients respond. A second signal would help. The METLUNG team in Reus, Spain ran multiplatform plasma metabolomics on 127 patients before their first immunotherapy dose and asked which circulating small molecules tracked clinical response, progression-free survival (PFS), and overall survival (OS). Five metabolites held up across all three endpoints.
Key Takeaways
Baseline plasma metabolomics can add prognostic information to clinical factors in NSCLC patients starting checkpoint inhibitors.
Higher blood histidine and citric acid tracked better immunotherapy outcomes in lung cancer; lactic acid, uracil, and sarcosine tracked worse ones.
In NSCLC immunotherapy patients, pairing clinical variables with metabolite data gave better early survival discrimination than either source alone.
Plasma metabolite signatures from METLUNG are prognostic, not treatment-predictive, and need multicenter validation before clinical use.
Multiplatform Metabolomics in a Real-World Immunotherapy Cohort
Treatment followed Spanish practice between 2020 and 2022, so the cohort was mixed: 39 patients had first-line immunotherapy alone, 47 had pembrolizumab with chemotherapy, and 41 started a checkpoint inhibitor after prior chemotherapy. Pre-dose EDTA plasma underwent targeted and semi-targeted metabolite profiling on GC-MS and LC-MS platforms, quantifying 575 metabolites against 163 reference standards. Stable disease counted as response. LASSO Cox models with nested cross-validation are a sensible guard when metabolites outnumber patients more than four to one, though no substitute for an external cohort.
Key Findings
Five metabolites survived every endpoint: Histidine, citric acid, uracil, lactic acid, and sarcosine stayed significant (FDR below 0.05) for response, PFS, and OS.
Histidine and citric acid as protective markers: Hazard ratios for PFS were 0.60 and 0.65, while uracil, lactic acid, and sarcosine sat between 1.33 and 1.36.
Combining data types worked best: The composite clinical-metabolomic model led on early time-dependent AUC and C-index; the five-metabolite model split risk groups with a PFS hazard ratio of 2.93.
The signal held without chemotherapy: In the immunotherapy-only subgroup, metabolomic signatures stratified PFS (HR 2.60) where the clinical model did not (HR 1.62, p=0.08).
Figure 1. Baseline plasma metabolites and immunotherapy outcomes in advanced NSCLC. Panel A shows volcano plots for clinical response (odds ratios) and for PFS and OS (hazard ratios), with metabolites linked to better outcomes in green and worse outcomes in red. Panel B pairs a heatmap of mean z-scored peak areas in responders and non-responders with forest plots for top-ranked metabolites, where citric acid and histidine fall on the protective side. Adapted from Lucía-Gozálvez et al. (2026), Journal for ImmunoTherapy of Cancer.
Reading the Hits Through Tumor Immunometabolism
Lower citric acid in non-responders is consistent with tumors shifting from TCA cycling toward glycolysis, and the matching rise in lactic acid describes that shift from the other end. Lactate is not inert. Brand and colleagues (2016) showed that LDHA-driven lactic acid blunts T and NK cell surveillance, which gives the association a plausible route to checkpoint resistance. Histidine is harder to explain, since its product histamine has been tied to suppressive macrophages in lung cancer, and the authors leave that question open.
What the Survival Models Can and Cannot Claim
Blood metabolite panels are multiplying in oncology, including a serum panel for early head and neck cancer, but this design limits the claims. Every patient received a checkpoint inhibitor with no comparator arm, so the markers are prognostic within treated patients rather than predictive of immunotherapy benefit. Discrimination for response was moderate (pooled AUC 0.72). With one center and no external set, these numbers are development-stage estimates.
In Practice: Handling Lactate-Sensitive Plasma Panels
Lactate is among the most handling-sensitive analytes in blood, because cells keep running glycolysis until plasma is separated. When we run targeted plasma metabolomics at Dalton for oncology programs, time to centrifugation and freeze-thaw history are logged per tube and checked against pooled QC drift before any marker is ranked. It's tedious bookkeeping, but it often separates a reproducible multi-omics biomarker discovery program from one whose top hit was really a courier delay.
Frequently Asked Questions
What is plasma metabolomics used for in cancer immunotherapy?
It measures hundreds of small molecules in a blood sample to capture the metabolic state of patient and tumor. In immunotherapy research it is used to find baseline markers linked to response or survival from a routine blood draw.
Which blood metabolites are linked to immunotherapy response in lung cancer?
In the METLUNG cohort, higher histidine and citric acid were associated with better outcomes on checkpoint inhibitors, while lactic acid, uracil, and sarcosine were associated with worse ones. These are single-center findings that still need external validation.
How much plasma does a targeted metabolomics study need?
Most targeted LC-MS or GC-MS assays use roughly 10 to 100 microliters of plasma, so a multiplatform panel may need a few hundred microliters per sample. Randomized run order and pooled QC samples matter as much as volume for keeping batch effects out of the results.
Conclusion
Pre-dose plasma does seem to carry prognostic information in checkpoint-treated NSCLC, and histidine and citric acid are reasonable candidates to carry forward. Whether any of them predicts benefit from immunotherapy itself, or survives another hospital's samples on another platform, isn't established. Teams planning immunotherapy biomarker work should bank tightly handled pre-dose plasma now so a targeted validation panel can be run later.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Lucía-Gozálvez, C., Papandreou, C., Badia, J., Cumeras, R., Peralta, S., Martí, C., Vidaller, A., Quintana, L., Martínez, S., & Gumà, J. (2026). METLUNG: Plasma metabolomic profiling identifies metabolomic signatures associated with immunotherapy outcomes in non-small cell lung cancer. Journal for ImmunoTherapy of Cancer, 14(9), e016077. https://doi.org/10.1136/jitc-2026-016077
Note
This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.
By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.