Early Post-Transplant AML Relapse, Caught on Surveillance: How Hard to Come Back
Caught early and at low burden, a post-transplant AML relapse doesn't obviously call for the most aggressive available option — but waiting has its own real cost if the gentler approach doesn't hold.
Walter B., a 65-year-old retired long-haul trucker, spent thirty years on interstate routes before settling into a slower life fixing small engines in his garage. He underwent a matched unrelated donor transplant for AML eight months ago — reduced-intensity conditioning, chosen then for his age and a modestly reduced baseline creatinine clearance, a decision that is about to matter a second time — and has felt entirely well since: no fevers, no new bruising, nothing that would have brought him in on his own. His relapse was caught on a routine day+240 surveillance marrow: 5% blasts, with donor chimerism drifting down from a stable 98% to 91% over the preceding six weeks. He is, by every symptom he can describe, feeling fine.
That asymptomatic, low-burden picture is exactly what makes the next step a genuine judgment call rather than an obvious one. Donor lymphocyte infusion works by reinfusing his original donor's T cells to reinforce the graft-versus-leukemia effect that helped control his disease the first time — a strategy with a real track record, though a lopsided one: DLI produces durable responses in the large majority of relapsed chronic myeloid leukemia, but only 15 to 30% of relapsed AML, a meaningfully weaker signal for his specific disease. A second full-intensity transplant is more definitive but carries a well-documented, substantially higher non-relapse mortality than a first, and would ask more of exactly the organ reserve that made his first conditioning reduced-intensity in the first place — a real cost to weigh against disease that, at 5% blasts, hasn't yet declared itself as aggressive relapse. His repeat molecular panel at relapse is unrevealing in a way that narrows rather than widens the options: no FLT3-ITD, nothing actionable, no targeted maintenance to layer on top of whatever is chosen. Azacitidine, a hypomethylating agent, offers a third path with a real mechanistic rationale beyond direct antileukemic effect: by demethylating and upregulating tumor antigen expression on residual leukemic cells, it can make them more visible to exactly the donor T cells a DLI would provide. Schroeder's prospective series combining the two specifically for post-transplant relapse reported responses meaningfully better than historical DLI alone, and reported them best in patients caught at low burden — which is the situation Walter's surveillance marrow has put him in, rather than a situation he would have to be manoeuvred into.
Responding to a relapse found before it announced itself
Post-transplant relapse, even caught early and at low burden, has historically carried a poor prognosis without definitive re-transplantation. Waiting on a strategy with a real chance of not working risks letting 5% blasts become overt relapse, at which point re-transplant becomes far less likely to succeed at all. I'd rather act definitively now, while his disease is still in the range where a second transplant has its best chance.
I agree the historical data on post-transplant relapse is sobering, but donor lymphocyte infusion alone in AML only produces durable responses in 15 to 30% of patients — genuinely weaker than its track record in chronic myeloid leukemia, where it works in the large majority. That's exactly why I wouldn't propose DLI alone. Azacitidine demethylates and upregulates tumor antigen expression on residual leukemic cells, making them more visible to donor T cells — Schroeder's series combining the two for post-transplant relapse reported response rates meaningfully better than DLI given by itself, with the best results in exactly the low-burden patients Walter resembles.
A second transplant carries a well-documented, substantially higher non-relapse mortality than a first — a real cost that shouldn't be paid before a lower-intensity option, with real supporting data at his disease burden, has had a fair chance.
I'd support the azacitidine-DLI combination, with one modification specific to Walter: he had mild skin GVHD at day 60, which tells us his immune system has already shown some alloreactive tendency even after resolving off immunosuppression. A standard, fixed-dose DLI in that setting carries real added GVHD risk. Escalating the donor T-cell dose stepwise — starting low and increasing only if there's no early GVHD flare — keeps the graft-versus-leukemia mechanism intact while respecting that his alloreactivity is not a blank slate.
I'd also want his second-transplant conditioning regimen and donor cells identified and held ready now, with an explicit response-check at two to three cycles, so we're not starting that search from zero if this approach doesn't hold.
Agreed: azacitidine combined with a graduated-dose donor lymphocyte infusion, response reassessed by marrow and chimerism at the end of cycle two. A second-transplant regimen and his original donor's cells are held on explicit standby rather than pursued only after this approach is judged to have failed.
Not fully agreed: how long is long enough to wait before declaring this approach unsuccessful. The transplant physician would prefer a shorter trial — one cycle, not two — before moving to second transplant, still concerned that his low burden today doesn't guarantee it stays low; the medical oncologist argued that combination responses in the literature often aren't visible until after a second cycle, and cutting the trial short risks abandoning a real chance prematurely. The two-cycle window was adopted as a compromise neither fully endorses as the ideal timeline for Walter specifically.