Hypothesis

Real‑time longitudinal liquid‑biopsy profiling of the tumor proteome and metabolome can forecast imminent therapeutic resistance by detecting a rise in glutamine‑derived α‑ketoglutarate (α‑KG) that fuels KDM5 histone demethylase activity. This metabolic‑epigenetic axis sustains a transcriptional program that drives both immune evasion and bypass signaling. Intervening early with a glutaminase inhibitor (or a cell‑permeable α‑KG antagonist) will break the loop, restore drug sensitivity, and prolong progression‑free survival.

Rationale

• Genomic alterations give a static blueprint but do not capture the functional state of the cell.[1]
• Proteomics reveals phosphorylation cascades and protein‑protein interactions that drive EMT and immune evasion.[2]
• Metabolomics shows shifts in glycolysis, TCA flux, and lipid remodeling that accompany therapy resistance.[2]
• Integrated proteogenomic‑metabolomic models already achieve AUC 0.81‑0.87 for early detection, proving the feasibility of real‑time readouts.[1]

Novel mechanistic link: Glutamine glutaminolysis produces α‑KG, a required cofactor for the lysine‑specific demethylase KDM5 family. Elevated KDM5 activity removes repressive H3K4me3 marks at promoters of genes involved in CXCR4 signaling, PD‑L1 expression, and ALDH1A3 stemness, thereby fostering a phenotype that survives targeted therapy and evades immune surveillance. This connection has not been explicitly tested in the context of adaptive liquid‑biopsy‑guided interventions.

Testable Predictions

1. Predictive biomarker: An increase in the ratio of plasma glutamate to α‑KG, coupled with a phosphoproteomic signature of heightened MAPK and AKT activity, will precede radiographic progression by ≥4 weeks in patients receiving standard targeted therapy.
2. Intervention effect: Administering a glutaminase inhibitor (e.g., CB‑839) at the first detection of this biomarker shift will reduce KDM5‑mediated demethylation (measurable by reduced H3K4me3 loss in circulating tumor‑derived nucleosomes) and delay the emergence of resistance‑associated protein isoforms.
3. Clinical outcome: The biomarker‑driven intervention arm will show a statistically significant improvement in median progression‑free survival (hazard ratio <0.75) compared with a control arm receiving therapy until radiographic progression.

Experimental Design (prospective N‑of‑1 platform)

• Cohort: 60 patients with advanced solid tumors (e.g., NSCLC, gastric cancer) on a standard targeted agent.
• Sampling: Plasma collected every two weeks for proteomics (TMT‑labelled LC‑MS/MS) and untargeted metabolomics (GC‑MS/LC‑MS).
• AI model: Trained on baseline and early‑on‑treatment multi‑omics to predict the glutamate/α‑KG rise and associated phosphoproteomic shift.
• Intervention trigger: When model probability exceeds 0.8, patients receive glutaminase inhibitor added to background therapy.
• Endpoints: Primary – PFS; Secondary – overall survival, change in circulating tumor nucleosome H3K4me3 levels, safety.

Falsifiability

If longitudinal sampling fails to show a consistent glutamate/α‑KG increase before resistance, or if adding glutaminase inhibition does not alter KDM5 activity or PFS, the hypothesis is refuted. Conversely, confirmation would validate a metabolically driven epigenetic resistance mechanism and establish a framework for real‑time, multi‑omics‑guided adaptive therapy.

References

• [1] https://pmc.ncbi.nlm.nih.gov/articles/PMC12634751/
• [2] https://www.metwarebio.com/integrative-proteomics-metabolomics-clinical-oncology/
• [3] https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2022.951137/full