DLBCL After CAR-T Failure: An Efficacy Debate That a Three-Hour Drive May Have Already Settled
A single patient with DLBCL relapsed after CAR-T therapy. The disagreement isn't only about which drug is more effective — it's about whether a real-world access barrier settles the question before the efficacy debate needs to.
B.F., a 68-year-old retired postmaster, relapsed with diffuse large B-cell lymphoma four months after achieving what her team had hoped would be a durable remission with axicabtagene ciloleucel, a CD19-directed CAR-T therapy given for her second relapse. Since the relapse was confirmed, her performance status has declined noticeably — she now tires walking from her car to the mailbox, something she managed easily even during her CAR-T recovery. She lives three hours from the academic center where her CAR-T was administered, in a small town whose local hospital does not have an oncology unit.
Glofitamab, a bispecific antibody bridging CD20 on the lymphoma to CD3 on her own T cells, was studied in the pivotal NP30179 expansion reported by Dickinson and colleagues, in which roughly a third of the 155 enrolled patients had already failed CAR-T — so she is inside the studied population rather than an extrapolation from it. The mechanistic worry that a T-cell engager should underperform in T cells exhausted by a prior T-cell therapy is a reasonable one, and that trial has actually answered it: complete response rates were 37% in patients with prior CAR-T and 39% in those without, which is no difference worth acting on. Where the concern survives is outside the trial, in real-world series of near-uniformly post-CAR-T patients, where response rates have run below the pivotal figures — a signal about selection and disease burden more than about her T cells. And underneath all of it sits something the efficacy argument cannot touch: glofitamab's cycle-one step-up schedule (obinutuzumab pretreatment, then 2.5mg and 10mg a week apart before the 30mg target dose) exists because cytokine release syndrome occurred in roughly 63% of trial patients and clustered around those first doses, which is why cycle one is monitored and usually inpatient. She lives three hours away, her performance status is ECOG 2 and falling, and her local hospital has no oncology unit.
Lymphoma tumor board, post-CAR-T relapse planning
Dickinson's pivotal NP30179 expansion enrolled 155 patients, about a third of them post-CAR-T, and reported a 39% complete response rate overall. That's direct evidence in her exact situation rather than an extrapolation, and the responses have proven durable — most complete responders in the extended follow-up were still in remission. I'd offer it as the primary recommendation.
I came in ready to argue T-cell exhaustion — CAR-T is itself a T-cell therapy, and a second T-cell engager asking the same exhausted compartment to work again is a real mechanistic worry. But the trial's own subgroup analysis addresses it directly: complete response was 37% with prior CAR-T against 39% without. I'm not going to lean on a mechanism the data has already looked at and not found.
What I'd hold onto instead is narrower and I think more honest: real-world cohorts of almost entirely post-CAR-T patients have reported lower response rates than the pivotal study, which is likelier to reflect who ends up in those cohorts — heavier pretreatment, bulkier disease, worse performance status — than anything about her T cells specifically. Her ECOG 2 and her four-month time to relapse put her closer to those cohorts than to the trial's median patient, and that's the discount I'd apply, not an immunologic one.
A positive pivotal trial result doesn't automatically mean the mechanism works equally well across every post-CAR-T patient it enrolled — exhaustion is a real, plausible reason some of that population responded less well, even if the trial as a whole looked favorable.
Before this gets resolved as a pure efficacy debate, I'd want to name something concrete: glofitamab's step-up dosing requires monitored, often hospital-based administration for at least the first cycle. She's three hours from the treating center, her performance status is declining, and her local hospital has no oncology unit. That's not a theoretical access concern — it may functionally take glofitamab off the table regardless of which of you turns out to be right about T-cell exhaustion.
Agreed: proceed with polatuzumab-based chemoimmunotherapy, deliverable closer to her home, rather than glofitamab.
Not agreed, and explicitly left unresolved: whether T-cell exhaustion genuinely blunts bispecific efficacy post-CAR-T — the clinical pharmacologist's concern was never actually tested in her case, since the logistics barrier settled the decision before the mechanistic question needed answering. The team was explicit that a patient with her exact disease and prior therapy but better geographic access might reasonably have received glofitamab first, leaving the exhaustion question for another case.