Hypothesis We propose that dynamic changes in phosphoproteomic signatures of metabolic enzymes, when coupled with real‑time shifts in specific lipid metabolites, can predict response to immune checkpoint blockade (ICB) in gastric cancer patients more accurately than static genomic or transcriptomic markers.

Mechanistic rationale

1. ICB efficacy depends on the ability of tumor cells to present antigens and sustain an inflamed microenvironment.
2. Phosphorylation of metabolic regulators such as PHGDH, ACLY, and ACSL1 redirects flux toward the hexosamine biosynthetic pathway and phospholipid remodeling, generating immunosuppressive lipids like ceramides and sphingosine‑1‑phosphate.
3. These lipid species modulate dendritic cell maturation and T‑cell infiltration, creating a feedback loop that either sustains or dampens immune activation.
4. Therefore, a combined read‑out of (a) site‑specific phosphorylation on metabolic enzymes and (b) the concentration of downstream immunosuppressive lipids captures the functional state that governs ICB sensitivity.

Testable predictions

• Patients exhibiting high basal phosphorylation of PHGDH‑Ser14 and elevated ceramide‑16:0 will show reduced progression‑free survival after anti‑PD‑1 therapy.
• Pharmacologic inhibition of the kinase driving PHGDH phosphorylation (e.g., CDK2) will decrease ceramide production and increase CD8⁺ T‑cell infiltration in pretreatment biopsies.
• Longitudinal sampling (baseline, on‑treatment day 8, and progression) will reveal that a decrease in phospho‑PHGDH concurrent with a drop in ceramide levels predicts durable response, whereas persistent signaling predicts early resistance.

Experimental design

• Cohort: 60 treatment‑naïve gastric adenocarcinoma patients slated for first‑line pembrolizumab.
• Sample collection: paired tumor biopsies at baseline and day 8 of therapy; plasma for lipidomics.
• Multi‑omics workflow:
  • Phosphoproteomics (TiO₂ enrichment, LC‑MS/MS) focusing on metabolic enzymes.
  • Targeted lipidomics (ceramides, sphingolipids) using MRM.
  • RNA‑seq for baseline transcriptional subtype (as control).
• Endpoints: objective response rate (RECIST v1.1), progression‑free survival, immune infiltrate (CD8⁺ IHC, multiplex immunofluorescence).
• Analysis:
  • Build a logistic regression model combining phospho‑site intensities and lipid concentrations.
  • Compare AUC against models based solely on mutational burden or PD‑L1 IHC.
  • Validate in an independent set of 30 patients.

Potential impact If confirmed, this phospho‑lipid signature could be implemented as a companion diagnostic, guiding early switch to combination regimens (e.g., ICB plus metabolic inhibitor) before radiographic progression, thereby addressing the current ~35% matching rate seen in genomics‑only trials 1.

Falsifiability A lack of correlation between the phospho‑lipid score and clinical outcome, or failure of metabolic inhibition to alter lipid levels and immune infiltration, would refute the hypothesis.