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

Abstract

Background:

Relapse is the main cause of treatment failure after allogeneic hematopoietic cell transplantation (allo-HCT). Therapeutic donor lymphocyte infusion (DLI) and second allo-HCT are common post-relapse options, but comparative evidence from the time of relapse is limited.

Methods:

We retrospectively analyzed 85 adults with acute leukemia or myelodysplastic syndromes (MDS) who received therapeutic DLI or a second allo-HCT after relapse (2010–2022). DLI required hematologic or clinical relapse. Overall survival (OS) was assessed using Kaplan–Meier analysis, Cox regression, and propensity score matching (PSM). Competing risks models evaluated relapse-related mortality (RRM) and non-relapse mortality (NRM).

Results:

60 patients received DLI and 25 underwent a second allo-HCT. No significant difference in OS was observed between groups (median OS 0.92 vs. 0.61 years, p=0.295). DLI showed higher RRM, while second allo-HCT showed a trend toward higher NRM. In the matched cohort (n = 36), OS remained numerically longer after second allo-HCT without significance.

Conclusion:

OS did not differ between second allo-HCT and therapeutic DLI. Treatment should be individualized, and larger multicenter studies are needed to refine patient selection.

Introduction

Relapse remains the most common cause of treatment failure following allogeneic hematopoietic cell transplantation (allo-HCT) in patients with acute leukemia and myelodysplastic syndrome (MDS) (1). Donor lymphocyte infusion (DLI) is a widely used post-relapse intervention intended to enhance the graft-versus-leukemia (GvL) effect (1–4). Overall survival (OS) after DLI was reported approximately 21% at 2 years, but outcomes improved when DLI is administered during complete remission (CR), with OS rates reaching up to 56% at 2 years (5). An alternative approach is a second allo-HCT for patients with relapse (6, 7). Patients who are eligible for a second allo-HCT are primarily those with good ECOG performance status, without evidence of graft-versus-host disease (GvHD) after first allo-HCT, and those who experienced a long remission period after first allo-HCT. Relapse after transplantation is associated with poor outcomes, with 2-year OS typically below 15% (8). A recent analysis found no significant OS difference between second allo-HCT and therapeutic DLI (9–11).

Advances in transplant techniques, broader donor availability, and expanded eligibility criteria have improved allo-HCT outcomes over the past decade (12, 13). HLA loss is a key relapse mechanism after haploidentical HCT, limiting DLI efficacy from the original donor (14). Changing donors at second allo-HCT offers no clear survival benefit (7, 15). Key prognostic factors influencing outcomes after relapse include the time to relapse and remission status prior to intervention (7, 11, 16).

Despite decades of clinical use, direct high-quality comparisons between therapeutic DLI and second allo-HCT are surprisingly scarce, leaving clinicians without clear guidance in this high-stakes setting. This uncertainty is especially relevant as both strategies carry substantial risks yet hold the potential for long-term survival in selected patients. In this study, we leverage a uniform, single-center cohort to provide a detailed, head-to-head comparison of these approaches, focusing on OS, RRM, and NRM. By applying robust statistical methods—including propensity score matching (PSM), restricted mean survival time (RMST), and competing risks regression—we aim to generate clinically relevant evidence that may inform decision-making in one of the most challenging scenarios in post-transplant care.

Methods

This retrospective, single-center cohort study was conducted in accordance with the Declaration of Helsinki and was approved by the local ethics committee (approval number 12/8/25). Adult patients (≥18 years) with acute leukemia or MDS who experienced relapse after a first allo-HCT between January 2010 and December 2022 were eligible. Patients were identified through the institutional transplant database and categorized according to their initial relapse-directed intervention (DLI) or second allo-HCT (see Figure 1).

Figure 1

This study focuses on patients who received active cellular therapy after relapse; a complete analysis of all relapsed patients was not feasible due to incomplete data capture and variable follow-up. Treatment selection between therapeutic DLI and second allo-HCT was based on clinical judgment within the institutional transplant program. In general, patients considered for second allo-HCT had favorable performance status, absence of uncontrolled GvHD, and were deemed fit for intensive treatment. In contrast, DLI was more commonly used in patients with advanced relapse, limited treatment tolerance, or when a second transplantation was not considered feasible. Treatment decisions were typically discussed within multidisciplinary transplant meetings.

Relapse was defined as the first documented evidence of recurrent or persistent disease after allo-HCT. In the absence of standardized confirmation of complete remission (CR) in all patients prior to intervention, this definition encompasses both true relapse and persistent disease. Disease monitoring was performed according to routine clinical practice and included serial assessment of donor chimerism, measurable residual disease (MRD), and, where available, morphologic evaluation. Donor chimerism and MRD assessments were routinely used for disease monitoring and could represent early indicators of disease recurrence prior to overt morphologic relapse.

Therapeutic DLI was defined as administration in the setting of hematologic relapse or clinical progression. Patients receiving DLI solely for declining donor chimerism or isolated MRD positivity without overt morphologic relapse were not included in this analysis, as these interventions were considered preemptive rather than therapeutic. Information on salvage therapy administered between relapse detection and therapeutic DLI was retrospectively collected and categorized as no salvage therapy, hypomethylating agent (HMA)-based therapy, HMA plus venetoclax, targeted therapy-based treatment, intensive salvage chemotherapy, or other approaches. By definition, therapeutic DLI in our cohort was administered in the setting of overt hematologic relapse; therefore, patients were considered to have active disease at the time of intervention. However, formal reassessment of remission status prior to DLI was not systematically performed, irrespective of intervening salvage therapy.

DLI were administered in the absence of active GvHD, typically at monthly intervals with stepwise dose escalation starting at 0.5 × 10⁶ CD3+ cells/kg and increasing to 5 × 10⁶, and up to 1 × 10⁷ cells/kg body weight, with a maximum of five infusions. The actual number of DLIs administered depended on donor lymphocyte availability, treatment tolerance, and clinical response. DLI administration was discontinued in cases of disease progression, treatment-related toxicity, or development of GvHD greater than grade I. In selected cases, DLI were combined with additional therapies such as hypomethylating agents according to clinical judgment.

The primary endpoint was OS, defined as the time from relapse until death from any cause, with survivors censored at last follow-up. Patients were analyzed according to their initial relapse-directed intervention (therapeutic DLI or second allo-HCT). Patients who later crossed over to a second allo-HCT after initial DLI remained in the DLI group for the primary analysis, as the aim was to evaluate outcomes from the time of relapse according to the initial treatment strategy. OS was analyzed using Kaplan–Meier methods and compared between groups with log-rank tests. Univariable and multivariable Cox proportional hazards models were used to assess prognostic factors. Baseline characteristics were compared between treatment groups using the Mann–Whitney U test for continuous variables and the chi-square or Fisher’s exact test for categorical variables, as appropriate. To reduce confounding, Nearest-neighbor PSM (1:1 ratio) was performed based on ECOG and time from first allo-HCT to relapse (17). Matching was performed on the logit of the propensity score using a greedy algorithm, without replacement, with a caliper of 0.2 SD of the logit (PS), on a complete-case basis. Covariate balance was evaluated using standardized mean differences, with |SMD| < 0.10 considered acceptable.

Fine–Gray competing risks models were used to assess relapse and non-relapse mortality (18). For the purpose of RRM analysis, any death occurring after documented relapse, persistent disease, or subsequent disease progression following intervention was classified as relapse-related mortality. This also included deaths indirectly attributable to uncontrolled malignancy, such as infectious complications occurring in the setting of active disease. NRM was defined as death occurring in the absence of documented relapse or disease progression. To account for the influence of long-term survivors, RMST was used to provide a robust measure in OS, as long-term survivors may disproportionately influence conventional survival estimates (19). To address potential immortal time bias, treatment exposure (therapeutic DLI and second allo-HCT) was additionally modeled as time-varying covariates using a Cox proportional hazards model. As a sensitivity analysis, a landmark analysis was performed at 60 days after relapse, including only patients who were alive and had received their respective intervention within the landmark period. In addition, a Simon–Makuch–type plot was generated to visualize time-dependent treatment effects over the course of follow-up. Detailed results of these analyses are provided in the Supplementary Tables. All statistical analyses were conducted using Python 3.11 (Python Software Foundation, https://www.python.org) with the packages pandas for data management, numpy for numerical computation, matplotlib for data visualization, lifelines for survival analysis, and scikit-learn/statsmodels for statistical modeling. Competing risk analyses were performed in R 4.5.1 (R Foundation for Statistical Computing, Vienna, Austria) using the cmprsk package.

Results

Patient characteristics

A total of 85 patients were analyzed, with 60 (71%) receiving therapeutic DLI and 25 (29%) undergoing a second allo-HCT. The median age at intervention was 60.1 years [IQR: 49.6–65.9] for the overall cohort. Age in the DLI group was 59.8 years [IQR: 49.8–65.7] and 53.4 years [IQR: 45.6–64.1] in the second allo-HCT. ECOG performance status was comparable across groups (median ECOG 1 [IQR: 0–1]), suggesting that most patients had good functional status at the time of intervention. The majority of patients were male (DLI: 56%, second allo-HCT: 60%). The most common relapsed disease was AML (50/85, 63%), followed by secondary AML (sAML; 19/85, 24%), MDS (11/85, 14%), and acute lymphoblastic leukemia (ALL; 5/85, 6%). Among 45 patients with available molecular and/or cytogenetic data at relapse, 31 (69%) were classified as high-risk and 14 (31%) as low-risk based on mutational profiles. Patients undergoing second allo-HCT had a significantly longer time from relapse to treatment compared to those receiving DLI (median 74.0 vs. 42.5 days, p<0.001). Other baseline characteristics, including diagnosis, conditioning intensity, time to relapse, donor type, and GvHD prior to relapse, did not differ significantly between groups.

Salvage therapy administered between relapse detection and cellular intervention differed between treatment groups. Patients proceeding to second allo-HCT more frequently received intensive cytoreductive chemotherapy regimens such as FLAG-based protocols, HAM, ICE, or fludarabine–cytarabine–based combinations. In contrast, patients treated with therapeutic DLI more often received lower-intensity approaches, including hypomethylating agent–based regimens with or without venetoclax, targeted therapies, or no prior salvage therapy. A detailed overview of salvage treatments is provided in Supplementary Table 1.

Among patients undergoing second allo-HCT, most received reduced-intensity conditioning (RIC) regimens (23/25, 92%), whereas myeloablative conditioning (MAC) was used in two patients (8%). Donor sources included matched unrelated donors in the majority of cases, with additional transplants from matched related, mismatched unrelated, and haploidentical donors.

Further clinical and genetic characteristics are summarized in Table 1. At the time of analysis, 70 patients (82%) had died. The primary causes of death were relapse or disease progression (n = 42, 49%), followed by sepsis (n = 20, 24%), GvHD (n = 2, 4%) and intracerebral bleeding (n = 2, 4%). The cause of death was unknown in four patients.

Median OS was 0.92 years (95% CI: 0.71–1.54) in the second allo-HCT group and 0.61 years (95% CI: 0.45–0.94) in the DLI group, with no statistically significant difference observed (log-rank p = 0.295; as shown in Figure 2). The OS plateau did not differ substantially between treatment groups, with 0.13 for second allo-HCT (after the last event at 5.91 years) and 0.12 for therapeutic DLI (after the last event at 3.46 years). In the PSM cohort (n = 36; 18 patients per group), median OS was longer in the second allo-HCT (1.01 years [95% CI: 0.60– 1.80]) compared to the DLI group (0.49 years [95% CI: 0.30–2.16]). However, the difference did not reach statistical significance (log-rank p = 0.519; as shown in Table 2 and Figure 3). Although deaths were numerically lower with DLI than second allo-HCT (15/18 vs. 13/18 with DLI), wide, overlapping CIs and a non-significant log-rank test suggest this difference is compatible with chance.

Figure 2

Figure 3

In a landmark analysis at 60 days after relapse (Supplementary Table 2), median OS from the landmark was 1.09 years (95% CI: 0.54–NR) in the second allo-HCT group (n=7) and 0.33 years (95% CI: 0.21–0.58) in the DLI group (n=39).

Predictors of overall survival

In a univariate Cox Regression, ECOG performance status at intervention was significantly associated with OS (HR = 1.51, 95% CI: 1.13–2.01; p = 0.005). Myeloablative conditioning during first allo-HCT was also associated with improved OS (HR = 0.58, 95% CI: 0.36–0.94; p = 0.027). Time from allo-HCT to relapse was also significantly associated with OS (HR = 0.87, 95% CI: 0.75-1.00, p=0.045). Other variables—including diagnosis of AML, age or prior GvHD—were not significantly associated with OS. In the multivariate model (Table 3), ECOG (HR =1.31, 95% CI: 0.98-1.79; p=0.072) showed a trend towards significance. Age, Conditioning intensity in first allo-HCT and time to relapse were not statistically significant predictors for OS. The RMST for 2 years was 0.95 years for patients receiving DLI and 1.04 years for those undergoing second allo-HCT with overlapping confidence intervals consistent with the absence of a statistically significant difference.

In time-dependent Cox analysis, both DLI and second allo-HCT were associated with increased mortality risk (DLI: HR 8.36, p=0.005; second allo-HCT: HR 6.27, p=0.018; Supplementary Table 3).

RRM versus NRM

To distinguish between RRM and NRM, a competing risk analysis was performed. At 1 year, the cumulative incidence of RRM was higher in patients receiving DLI (43%) compared to those undergoing a second allo-HCT (24%). In contrast, the incidence of NRM was lower in the DLI group (20% vs. 34%). A similar pattern was observed at 2 years, with RRM reaching 50% in the DLI group and 38% in the second allo-HCT group, while NRM increased to 24% and 40%, respectively (Table 4, Figure 4). Fine–Gray regression confirmed that second allo-HCT was associated with a lower subdistribution hazard (sHR) for RRM compared to DLI (sHR = 0.63, 95% CI: 0.34–1.17, p = 0.14), consistent with the observed cumulative incidence functions (CIF). Conversely, a trend toward increased NRM was observed in the second allo-HCT group (sHR = 1.81, 95% CI: 0.82–4.01, p = 0.14).

Figure 4

Competing risk analysis comparing the cumulative incidence of RRM and NRM between DLI and second allo-HCT. 1-year and 2-year incidences are indicated. Events were mutually exclusive, with death without prior relapse defined as NRM.

Discussion

After relapse post–allo-HCT, clinicians must choose between therapeutic DLI, a lower-intensity option with modest efficacy in overt relapse, and a second allo-HCT, which may improve disease control at the cost of toxicity. In our full cohort, competing-risks (Fine–Gray) regression showed a lower sHR of relapse-related mortality with second allo-HCT alongside a non-significant trend toward higher non-relapse mortality. In the propensity-score–matched subset (n = 36), we did not detect a difference in OS, although OS and RMST were numerically longer after second allo-HCT.

ECOG performance status at first allo-HCT showed a trend towards significance for OS, plausibly reflecting baseline fitness and selection for more intensive regimens; prior reports in the second-transplant setting support the prognostic value of ECOG (20, 21). Furthermore, an EBMT registry analysis found no significant association between conditioning intensity—whether applied during first or second allo-HCT —and OS (16).

A key methodological distinction of our study is the choice of relapse as time zero, rather than the date of intervention. This avoids immortal-time bias from variable delays to treatment and provides a more realistic, ‘real-world’ estimate of outcomes. Methodologically, we summarized disease-specific risks using competing-risks regression in the full cohort and compared OS/RMST in a PSM subset. These approaches reduce overt confounding and complement median-based summaries, but several limits temper inference: the matched sample was small (n = 36); PSM was restricted to a limited number of variables due to sample size and imbalance in baseline characteristics, particularly with respect to disease type, precluding full adjustment for all clinically relevant confounders; post-relapse salvage after second allo-HCT (DLI in six; a third transplant in one) may have influenced cumulative incidences; and censoring at crossover—though symmetric—can still be informative if eligibility for further therapy tracks with prognosis. Heterogeneity in disease biology and prior treatment further restricts generalizability. Accordingly, all between-group contrasts should be viewed as exploratory. To further address potential immortal-time bias, we performed additional time-dependent Cox regression and landmark analyses, which yielded results consistent with the primary findings.

Differences in salvage therapy prior to cellular intervention may also have influenced outcomes. Patients undergoing second allo-HCT frequently received intensive cytoreductive chemotherapy before transplantation, whereas DLI was often administered after lower-intensity treatments or without prior salvage therapy. This reflects common clinical practice but may have contributed to differences in disease burden at the time of cellular therapy. Previous studies demonstrated superior outcomes when DLI is administered in the setting of molecular relapse, MRD positivity, or low disease burden compared to overt morphologic relapse (22–24). These findings suggest that disease state at the time of intervention substantially influences the efficacy of relapse-directed cellular therapy. However, the present study intentionally focused on therapeutic DLI administered in patients with overt hematologic relapse, whereas preemptive DLI strategies were excluded from the analysis. In contrast, some patients undergoing second allo-HCT achieved remission following salvage therapy prior to transplantation, which may have influenced outcome comparisons between treatment groups.

In this cohort, we were unable to detect a significant OS difference between second allo-HCT and therapeutic DLI. The opposing signals-lower RRM and higher NRM with second allo-HCT—did not translate into improved OS, and the OS prognosis after relapse post allo-HCT remained poor, consistent with prior series (5, 9, 11, 25–27). Thus, no OS-based preference can be recommended from our data; in practice, treatment selection will continue to be driven by goals of care, toxicity constraints, and feasibility rather than an expectation of OS gain.

The cumulative DLI dose exposure may have contributed to the modest efficacy, reflecting our institutional practice of stepwise dose escalation to minimize GvHD risk. In addition, OS after relapse was short in many patients, limiting the time window for repeated DLI administrations. Five of 60 patients received five DLIs, while 14 patients received four infusions. The total number of DLIs delivered was therefore often constrained by early disease progression or clinical deterioration. In addition, the cumulative DLI dose also depended on donor lymphocyte availability, as the number of cells obtained from donor collections varied between donors.

Larger multicenter cohorts using PSM and prespecified salvage/crossover handling should test whether any OS advantage exists in clearly defined clinical subsets and refine patient-selection frameworks.

Limitations

This study has several limitations. First, its retrospective, single-center design may have introduced selection bias and limited the generalizability of our findings. Although PSM was used to adjust for baseline differences between treatment groups, residual confounding due to unmeasured variables cannot be ruled out. Second, the overall cohort size was modest, particularly within the matched subgroups, thus limiting the statistical power to detect small but clinically relevant survival differences. Third, molecular and cytogenetic data at relapse were incomplete in a subset of patients, which may have affected risk stratification and restricted more granular analyses of disease biology. In addition, the cohort included biologically heterogeneous disease entities (AML, sAML, MDS, ALL), with MDS exclusively represented in the DLI group, resulting in a baseline imbalance and potential residual confounding. Due to limited sample size, disease-specific subgroup analyses were not performed. Moreover, because this study was restricted to patients receiving active cellular relapse-directed therapy after overt relapse, selection bias cannot be excluded and the findings are not directly generalizable to the entire population of patients relapsing after first allo-HCT. In addition, due to the lack of systematic assessment of remission status prior to intervention, the distinction between true relapse and persistent disease could not be reliably established, which may have introduced some degree of misclassification. Lastly, the study covers a relatively long time period (2010–2022), during which supportive care, DLI strategies, and transplantation protocols have evolved—potentially influencing outcomes independently of the relapse intervention.

Conclusion

We did not detect an OS difference between second allo-HCT and therapeutic DLI, including after PSM for key confounders (e.g., ECOG, time to relapse). Given the intensity and toxicity of second allo-HCT, treatment selection should remain individualized, taking into account performance status, remission duration, and patient preferences, as no OS benefit was observed for either strategy. Larger, prospective multicenter studies are warranted to validate these findings and to determine whether an OS advantage persists when analyzed from the time of relapse, thereby eliminating immortal time bias.

References

• 1

  TsirigotisPByrneMSchmidCBaronFCiceriFEsteveJet al. Relapse of AML after hematopoietic stem cell transplantation: Methods of monitoring and preventive strategies. A review from the ALWP of the EBMT. Bone Marrow Transplant. (2016) 51:1431–8. doi: 10.1038/bmt.2016.167

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  MandigersCMPWVerdonckLFMeijerinkJPPDekkerAWSchattenbergAVMBRaemaekersJMM. Graft-versus-lymphoma effect of donor lymphocyte infusion in indolent lymphomas relapsed after allogeneic stem cell transplantation. Bone Marrow Transplant. (2003) 32:1159–63. doi: 10.1038/sj.bmt.1704290

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  MaurerKAntinJH. The graft versus leukemia effect: donor lymphocyte infusions and cellular therapy. Front Immunol. (2024) 15:1328858. doi: 10.3389/fimmu.2024.1328858

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  BiederstädtARezvaniK. How I treat high-risk acute myeloid leukemia using preemptive adoptive cellular immunotherapy (2023) 11:812207. Available online at: http://ashpublications.org/blood/article-pdf/141/1/22/2073317/blood_bld-2021-012411-c-main.pdf (Accessed June 1, 2026).

  • Google Scholar
  • View reference in article

• 5

  SchmidCLabopinMNaglerABornhäuserMFinkeJFassasAet al. Donor lymphocyte infusion in the treatment of first hematological relapse after allogeneic stem-cell transplantation in adults with acute myeloid leukemia: A retrospective risk factors analysis and comparison with other strategies by the EBMT acute leukemia working party. J Clin Oncol. (2007) 25:4938–45. doi: 10.1200/JCO.2007.11.6053

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  GyurkoczaBStorbRChaunceyTRMaloneyDGStorerBESandmaierBM. Second allogeneic hematopoietic cell transplantation for relapse after first allografts. Leuk Lymphoma. (2019) 60:1758–66. doi: 10.1080/10428194.2018.1542149

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  Kharfan-DabajaMAReljicTYassineFNishihoriTKumarATawkMMet al. Efficacy of a second allogeneic hematopoietic cell transplant in relapsed acute myeloid leukemia: results of a systematic review and meta-analysis. Transplant Cell Ther. (2022) 28:767.e1–767.e11. doi: 10.1016/j.jtct.2022.08.008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  WebsterJALuznikLGojoI. Treatment of AML relapse after allo-HCT. Front Oncol. (2021) 11:812207. doi: 10.3389/fonc.2021.812207

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  OrtíGSanzJGarcía-CadenasISánchez-OrtegaIAlonsoLJiménezMJet al. Analysis of relapse after transplantation in acute leukemia: A comparative on second allogeneic hematopoietic cell transplantation and donor lymphocyte infusions. Exp Hematol. (2018) 62:24–32. doi: 10.1016/j.exphem.2018.03.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  YanadaMKonumaTYamasakiSKondoTFukudaTShingaiNet al. Relapse of acute myeloid leukemia after allogeneic hematopoietic cell transplantation: clinical features and outcomes. Bone Marrow Transplant. (2021) 56:1126–33. doi: 10.1038/s41409-020-01163-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  Kharfan-DabajaMALabopinMPolgeENishihoriTBazarbachiAFinkeJet al. Association of second allogeneic hematopoietic cell transplant vs donor lymphocyte infusion with overall survival in patients with acute myeloid leukemia relapse. JAMA Oncol. (2018) 4(9):1245–53. doi: 10.1001/jamaoncol.2018.2091

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  SchmidCSchmaelterAKNgoyaMGalimardJEBazarbachiAFinkeJet al. Continuously improving outcome over time after second allogeneic stem cell transplantation in relapsed acute myeloid leukemia - an EBMT registry analysis of 1540 patients (2024). Available online at: https://www.researchsquare.com/article/rs-3996844/v1.

  • Google Scholar
  • View reference in article

• 13

  GooleyTAChienJWPergamSAHingoraniSSorrorMLBoeckhMet al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Engl J Med. (2010) 363:2091–101. doi: 10.1056/NEJMoa1004383

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  VagoLPernaSKZanussiMMazziBBarlassinaCStanghelliniMTLet al. Loss of mismatched HLA in leukemia after stem-cell transplantation. N Engl J Med. (2009) 361:478–88. doi: 10.1056/NEJMoa0811036

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  ChristopeitMKussOFinkeJBacherUBeelenDWBornhäuserMet al. Second allograft for hematologic relapse of acute leukemia after first allogeneic stem-cell transplantation from related and unrelated donors: The role of donor change. J Clin Oncol. (2013) 31(26):3259–71. doi: 10.1200/JCO.2012.44.7961

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  RuutuTDe WreedeLCVan BiezenABrandRMohtyMDregerPet al. Second allogeneic transplantation for relapse of Malignant disease: Retrospective analysis of outcome and predictive factors by the EBMT. Bone Marrow Transplant. (2015) 50:1542–50. doi: 10.1038/bmt.2015.186

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  AustinPC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. (2011) 46:399–424. doi: 10.1080/00273171.2011.568786

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  FineJPGrayRJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. (1999) 94(446):496–509. doi: 10.1080/01621459.1999.10474144

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  RoystonPParmarMK. Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome. BMC Med Res Methodol. (2013) 13:152. doi: 10.1186/1471-2288-13-152

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  SchneidawindCHagmaierVFaulCKanzLBethgeWSchneidawindD. Second allogeneic hematopoietic cell transplantation enables long-term disease-free survival in relapsed acute leukemia. Ann Hematol. (2018) 97:2491–500. doi: 10.1007/s00277-018-3454-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  VrhovacRLabopinMCiceriFFinkeJHollerETischerJet al. Second reduced intensity conditioning allogeneic transplant as a rescue strategy for acute leukaemia patients who relapse after an initial RIC allogeneic transplantation: Analysis of risk factors and treatment outcomes. Bone Marrow Transplant. (2016) 51:186–93. doi: 10.1038/bmt.2015.221

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  KrishnamurthyPPotterVTBarberLDKulasekararajAGLimZYPearceRMet al. Outcome of donor lymphocyte infusion after T cell-depleted allogeneic hematopoietic stem cell transplantation for acute myelogenous leukemia and myelodysplastic syndromes. Biol Blood Marrow Transplant. (2013) 19:562–8. doi: 10.1016/j.bbmt.2012.12.013

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  TanYDuKLuoYShiJCaoLZhengYet al. Superiority of preemptive donor lymphocyte infusion based on minimal residual disease in acute leukemia patients after allogeneic hematopoietic stem cell transplantation. Transfusion (Paris). (2014) 54:1493–500. doi: 10.1111/trf.12524

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  SchmidCLabopinMSchaapNVeelkenHBrechtAStadlerMet al. Long-term results and GvHD after prophylactic and preemptive donor lymphocyte infusion after allogeneic stem cell transplantation for acute leukemia. Bone Marrow Transplant. (2022) 57:215–23. doi: 10.1038/s41409-021-01515-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  Al-ShaibaniEBautistaRLiptonJHKimDDViswabandyaAKumarRet al. Comparison of outcomes after second allogeneic hematopoietic cell transplantation versus donor lymphocyte infusion in allogeneic hematopoietic cell transplant patients. Clin Lymphoma Myeloma Leuk. (2022) 22:e327–34. doi: 10.1016/j.clml.2021.11.004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  BejanyanNWeisdorfDJLoganBRWangHLDevineSMde LimaMet al. Survival of patients with acute myeloid leukemia relapsing after allogeneic hematopoietic cell transplantation: A center for international blood and marrow transplant research study. Biol Blood Marrow Transplant. (2015) 21:454–9. doi: 10.1016/j.bbmt.2014.11.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 27

  ChangYJHuangXJ. Donor lymphocyte infusions for relapse after allogeneic transplantation. When, if and for whom? Blood Rev. (2013) 27:55–62. doi: 10.1016/j.blre.2012.11.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article