Skip to content

UPenn intracerebroventricular bivalent CAR-T (EGFR + IL-13Rα2) for recurrent GBM

Ph1 update of locoregional bivalent EGFR + IL-13Rα2 CAR-T delivered into the CSF via Ommaya for adult recurrent glioblastoma.

  • Speaker(s): Stephen J. Bagley, MD, MSCE; co-investigators Donald M. O'Rourke, Carl H. June (lab and clinical team)
  • Affiliation: Abramson Cancer Center / Perelman School of Medicine, University of Pennsylvania
  • Anchor type: cell therapy / trial
  • Day / session: Day 1 (Mon Mar 23, 2026), Cellular Therapy for GBM session; Bagley also co-chaired the meeting
  • Status: post-meeting; details captured from AACR Brain 26 program notes, Penn Medicine institutional roundup, and the published Nature Medicine 2024 / PMC 2025 dataset that the cohort expansion extends; specific Brain 26 cohort numerics TBD pending Cancer Research abstract supplement

Headline finding

Updated Ph1 cohort of locoregionally delivered, bicistronic CAR-T co-targeting EGFR and IL-13Rα2 in adult recurrent GBM, building on the Nature Medicine 2024 first-in-human report that documented rapid, transient tumor regressions in the index patients. The 2026 update reads out an expanded, ICV-dosed (intracerebroventricular, via Ommaya reservoir) cohort with longer follow-up. Exact ORR, durability, and DLT numbers from the AACR Brain 26 talk TBD pending the abstract supplement.

Mechanism / methodology

A single bicistronic CAR construct co-expresses two scFvs — one against EGFR (wild-type and variant III) and one against IL-13Rα2 — paired to a 4-1BB costimulatory domain (per the Nature Medicine 2024 construct). Cells are delivered into the CSF via Ommaya reservoir rather than IV or intratumoral, exploiting CSF circulation to reach multifocal disease and limit on-target/off-tumor systemic EGFR toxicity. Dual antigen targeting is designed to blunt single-target antigen escape seen with EGFRvIII-only CAR-T.

Mechanism / methodology deep dive

Underlying biology. EGFR (wild-type and the tumor-specific variant III deletion) and IL-13Rα2 are two of the most consistently overexpressed surface antigens on adult GBM, with largely non-overlapping expression patterns across single-cell-defined tumor cell states (EGFR skews toward proneural/OPC-like; IL-13Rα2 toward mesenchymal). Single-target EGFRvIII CAR-T (Penn 2017, Nature Translational Medicine) cleared antigen-positive cells but left an antigen-negative reservoir, motivating the bivalent design.

Therapeutic design. A single bicistronic lentiviral construct co-expresses an EGFR-directed scFv (the Penn group's anti-EGFR(806)-class binder, recognizing both wild-type and vIII conformational epitope) and an IL-13Rα2 binder, each linked to a 4-1BB / CD3ζ signaling cassette. ICV delivery uses an Ommaya reservoir; single or limited-dose schedule reported in the Nature Medicine 2024 first-in-human; lymphodepletion regimen and re-dose schedule for the expansion cohort TBD.

Resistance / failure modes. Documented: rapid cytokine-driven neuroinflammation requiring tocilizumab/anakinra; transient response with re-growth, mechanism not yet defined but candidates include (i) loss of both antigens via cell-state switch, (ii) T-cell exhaustion in the CSF/TME, and (iii) intrinsic resistance in NPC-like / proliferative-stem subpopulations.

Translational gaps. Re-dosing cadence via Ommaya, optimal lymphodepletion for CSF delivery, and whether bivalent ICV is genuinely additive over single-target IL-13Rα2 ICV in a matched cohort — none publicly defined. TBD whether CAR-T persistence in CSF correlates with durability.

Where it fits in the corpus

  • AACR 2026 (Annual): Bagley / O'Rourke / June group also presented neuro-onc cell-therapy sessions in San Diego four weeks later.
  • SITC 2026: solid-tumor CAR-T track; UPenn ICV bivalent is the lead adult-GBM design benchmark.
  • Related Brain 26 vault entries: city-of-hope-il13ra2-cart.md, monje-gd2-cart-dipg.md, neuroscience-of-gliomas-tam.md.

Open questions

  • How does durability of the bivalent ICV approach compare to single-target IL-13Rα2 locoregional (City of Hope) in matched recurrent GBM cohorts, and does dual targeting actually reduce antigen-escape relapse?
  • Optimal lymphodepletion and re-dosing schedule via Ommaya — single dose vs serial intraventricular infusions — is not yet defined publicly.