Hypothesis

HER2‑low breast and gastric cancers display a metabolic shift that increases reliance on glutamine‑derived α‑ketoglutarate to sustain TCA cycle activity when HER2 signaling's dampened. This glutamine dependence creates a synthetic lethal interaction with glutaminase inhibition, which can be anticipated by measuring plasma levels of glutamate, glutamine and downstream TCA intermediates.

Mechanistic Rationale

Multi‑omics analyses show that ERBB2/HER2 expression inversely correlates with glycolysis and alanine‑aspartate‑glutamate pathways in tumor tissue and bloodERBB2/HER2 negatively correlates with immune infiltration and AAG/GG metabolic pathways, with metabolomic blood validation. When HER2 signaling's low, cells compensate by upregulating amino acid transporters (e.g., SLC1A5) and GLS to funnel glutamine into the TCA cycle via α‑ketoglutarate, thereby maintaining ATP production and redox balanceMulti-omics integration captures functional and phenotypic cancer traits beyond genetic alterations. This compensatory flux is not captured by genomics alone but emerges in proteogenomic and metabolomic layers, especially in the phosphorylation status of GLS and the acetylation of mitochondrial enzymes. AI‑driven fusion models that prioritize genomics for therapy prediction may miss this metabolic vulnerability because they underweight metabolomic featuresMulti-omics integration captures functional and phenotypic cancer traits beyond genetic alterations.

Testable Predictions

1. Patient‑derived xenografts (PDX) with low HER2 protein but high GLS expression 'll show greater tumor growth inhibition when treated with a HER2 antibody‑drug conjugate plus a GLS inhibitor than with either agent alone.
2. In a prospective cohort of HER2‑low metastatic patients, baseline plasma glutamate/glutamine ratio > 1.5 'll predict progression‑free survival < 6 months on HER2‑targeted monotherapy.
3. CRISPR‑mediated knockdown of GLS in HER2‑low cell lines 'll restore sensitivity to HER2 inhibition and reduce intracellular α‑ketoglutarate levels, measurable by targeted metabolomics.

Experimental Design

• In vitro: We'll measure GLS protein, phospho‑GLS, and SEAHORSE OCR/ECAR after glutamine withdrawal. We'll treat with trastuzumab deruxtecan, CB‑839 (GLS inhibitor), and combination. We'll assess viability, apoptosis, and metabolomic shifts via LC‑MS.
• In vivo: We'll generate PDX models stratified by HER2 IHC and GLS immunohistochemistry. We'll randomize to control, HER2‑targeted monotherapy, GLS inhibitor monotherapy, or combination. We'll track tumor volume, we'll perform serial blood draws for targeted metabolomics (glutamate, glutamine, α‑KG, succinate). We'll correlate metabolic changes with tumor response.
• Clinical: We'll partner with an ongoing HER2‑low basket trial. We'll collect pretreatment plasma, we'll run targeted metabolomics, and we'll stratify patients by glutamate/glutamine ratio. Primary endpoint: PFS at 6 months. Secondary: objective response rate, metabolic response.

Potential Impact

If it's validated, this hypothesis would provide a mechanistic explanation for why some HER2‑low tumors resist HER2‑directed therapy and offer a concrete, blood‑based biomarker to guide combination treatment with glutaminase inhibitors. It also demonstrates how moving beyond genomics‑centric AI models to incorporate dynamic metabolomic signals can improve therapeutic resistance prediction, addressing a key gap highlighted in recent multi‑omics reviewsMulti-omics integration captures functional and phenotypic cancer traits beyond genetic alterations.