The research team, led by Alexander C. Angleitner and colleagues, analyzed 85 adults with acute leukemia or MDS who relapsed after a first allogeneic hematopoietic cell transplantation between 2010 and 2022. Sixty patients received therapeutic donor lymphocyte infusions (DLI), while 25 underwent a second allogeneic transplant. Unlike many prior comparisons, the researchers anchored their analysis at the moment of relapse rather than the date of treatment, a methodological choice designed to eliminate so-called immortal time bias, in which patients who survive long enough to receive a delayed intervention artificially inflate that intervention’s apparent benefit.

The central finding was stark in its symmetry. Median overall survival from relapse was 0.92 years for patients who received a second transplant and 0.61 years for those treated with DLI, a difference that failed to reach statistical significance (p = 0.295). Even after propensity score matching, which paired patients on performance status and time to relapse to reduce confounding, the matched cohort of 36 patients showed a numerically longer median survival after second transplantation (1.01 years versus 0.49 years), but the difference remained statistically indistinguishable from chance.

Beneath the surface of equivalent survival, however, the two strategies revealed opposing risk profiles. Using Fine-Gray competing risks regression, the researchers found that patients treated with DLI faced a higher cumulative incidence of relapse-related mortality, reaching 50 percent at two years compared with 38 percent after second transplantation. Conversely, non-relapse mortality, deaths from treatment complications rather than the disease itself, trended higher after second transplantation, reaching 40 percent at two years versus 24 percent after DLI. Neither contrast achieved statistical significance, but the pattern is clinically meaningful: the second transplant appears to control disease more effectively while exacting a heavier toll from toxicity.

The biology behind these divergent risks reflects the fundamental nature of each intervention. DLI works by harnessing the graft-versus-leukemia effect, in which donor T cells recognize and attack malignant cells. But its efficacy drops sharply when disease burden is high. Prior studies have shown that DLI administered at the molecular relapse stage or for minimal residual disease positivity can yield two-year survival rates approaching 56 percent, whereas DLI given during overt hematologic relapse produces survival around 21 percent at two years. The Göttingen study deliberately focused on therapeutic DLI, meaning infusions given only after documented hematologic or clinical relapse, excluding preemptive use for falling donor chimerism or isolated MRD positivity.

DLI protocols in the study followed a stepwise dose-escalation scheme, beginning at 0.5 million CD3-positive T cells per kilogram of body weight and escalating up to 10 million cells per kilogram across as many as five monthly infusions. Infusions were halted in cases of disease progression, unacceptable toxicity, or graft-versus-host disease exceeding grade I. The median patient received three infusions. The authors note that cumulative dose exposure may have been constrained by early disease progression and by variable donor lymphocyte availability, potentially limiting the therapy’s efficacy in this overt-relapse setting.

Second transplantation, by contrast, offered more aggressive disease control at the cost of intensity. Ninety-two percent of second transplants used reduced-intensity conditioning regimens, and 96 percent involved a different donor than the first transplant, though prior evidence suggests donor change itself confers no clear survival advantage. Notably, patients heading toward a second transplant waited longer from relapse to treatment, a median of 74 days versus 42.5 days for DLI, largely because many underwent intensive cytoreductive chemotherapy first, including FLAG, HAM, ICE, or fludarabine-cytarabine protocols. DLI recipients more often received lower-intensity bridging such as hypomethylating agents with or without venetoclax, or no salvage therapy at all.

Prognostic factors emerged clearly from the analysis. Performance status at intervention was the strongest predictor of survival, with each increment in ECOG score associated with a 51 percent increase in the hazard of death in univariate analysis. Myeloablative conditioning during the first transplant and longer time from transplant to relapse were also associated with better outcomes, both well-established markers of more indolent disease biology. Restricted mean survival time analysis, which captures average survival over a fixed horizon and is less sensitive to long-term survivor outliers, showed 0.95 years of expected survival over two years for DLI patients versus 1.04 years for second transplant recipients, again with overlapping confidence intervals.

The study’s limitations are substantial and acknowledged candidly by the authors. The single-center, retrospective design introduces selection bias: patients chosen for second transplantation generally had better functional status, no uncontrolled graft-versus-host disease, and longer remissions, while DLI was more often deployed in advanced relapse or when transplantation was deemed infeasible. The matched cohort of 36 patients was too small to detect modest survival differences, molecular and cytogenetic data were missing in nearly half the cohort, and MDS patients appeared exclusively in the DLI group, creating baseline imbalance. Crossover was also common, with 15 percent of DLI patients eventually receiving a second transplant and 24 percent of second-transplant patients later receiving DLI.

Despite these caveats, the study carries a clear message for clinicians and patients navigating relapse after transplant. With two-year survival after post-transplant relapse typically below 15 percent in historical series, and no survival advantage demonstrated for either strategy in this matched comparison, treatment selection should remain individualized, driven by performance status, remission duration, patient preferences, and feasibility rather than by any expectation of a survival gain. The authors call for larger, prospective multicenter studies with prespecified handling of salvage therapy and crossover to determine whether defined patient subsets might benefit preferentially from one approach. Until such data exist, the choice between a second transplant and therapeutic DLI remains a judgment call, made at the bedside rather than dictated by the evidence.

Subject of Research: Comparing therapeutic donor lymphocyte infusion and second allogeneic transplantation for relapse after allogeneic stem cell transplantation in acute leukemia and MDS

Article Title: Matched comparison of therapeutic DLI and second allogeneic transplantation after relapse of acute leukemia or MDS

Article References: Angleitner, A. C., Maulhardt, M., Hasenkamp, J., Büntzel, J., & Wulf, G. (2026). Matched comparison of therapeutic DLI and second allogeneic transplantation after relapse of acute leukemia or MDS. Frontiers in Hematology, 5, Article 1826416. https://doi.org/10.3389/frhem.2026.1826416

Image Credits: AI Generated

DOI: 10.3389/frhem.2026.1826416

Keywords: donor lymphocyte infusion, second allogeneic transplantation, acute myeloid leukemia, myelodysplastic syndrome, post-transplant relapse, graft-versus-leukemia, overall survival, non-relapse mortality, relapse-related mortality, propensity score matching, hematopoietic cell transplantation, competing risks analysis