Every one of six lymphoma patients carrying a broken STXBP2 gene developed toxicity after treatment with Kite Pharma’s Yescarta, according to a new genomic analysis of Kite’s trial data published today in Science Immunology. Zero exceptions.
Mark Leick, an oncologist at Massachusetts General Hospital, led the study out of Marcela Maus’s lab, sequencing germline DNA from patients treated in Kite’s ZUMA-1 and ZUMA-7 trials, the data that got axicabtagene ciloleucel approved for refractory large B-cell lymphoma. The STXBP2 signal only showed up in ZUMA-1’s sicker population, not ZUMA-7. Leick attributes the gap to how inflamed those patients already were, not a false result. The team also flagged ADAMTSL3 variants tied to lower toxicity, and PTPN22 variants linked to how aggressively CAR-T cells multiply after infusion, a pattern confirmed in both trials.
Carl June, the Penn researcher who built the first CAR-T, called the PTPN22 finding “particularly striking” and said knocking the gene out during manufacturing could boost cell expansion and cut required doses.
That’s the real stakes here. If germline screening becomes a manufacturing input, every CAR-T slot picks up a new cost and a new timeline before apheresis even starts. It also raises a liability question the field hasn’t had to answer yet: what happens when a patient’s own DNA predicted the toxicity and nobody checked.
One catch limits how far this goes right now. Leick’s cohort was almost entirely patients of European descent — there weren’t enough non-European patients enrolled in Kite’s trials to run the analysis. Leick says he’d love to run it again once that changes.
Sarah Chen