Donor Lymphocyte Infusion in the Treatment of Post-Transplant Relapse of Acute Myeloid Leukemias and Myelodysplastic Syndromes Significantly Improves Overall Survival: A French–Italian Experience of 134 Patients

Simple Summary Allogeneic stem cell transplantation (allo-SCT) represents the only potentially curative treatment for high-risk acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS), but up to 50% of patients relapse after allo-SCT. The salvage therapy after disease recurrence is not standardized, and the outcome remains unfavorable. Therefore, there is a growing interest in determining the most effective approach to manage the post-transplant phase with the goal of promptly detecting disease recurrence or preventing it. In this context, we conducted a retrospective study to assess the overall survival (OS) of patients with relapsed AML or MDS after allo-SCT with the aim of acquiring useful information for identifying the best prospective therapeutic strategy. The OS was evaluated according to the type of therapy, whether it included donor lymphocyte infusion (DLI), the timing of administration, and whether it occurred during an overt hematological relapse or in a preemptive setting. Abstract Background: Disease relapse after allogeneic stem cell transplantation (allo-SCT) is the main challenge for curing acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). We investigated the overall survival (OS) after allo-SCT relapse according to different therapeutic approaches. Methods: We analyzed 134 patients who relapsed after allo-SCT performed between 2015 and 2021 at Saint-Antoine University Hospital, Paris and Spedali Civili di Brescia, Brescia. Of these, 103 (77%) were treated, comprising 69/103 (67%) who received therapy in overt relapse and 34/103 (33%) who were treated in a pre-emptive manner when molecular/cytogenetics recurrence or mixed chimerism occurred. The treatment was donor lymphocyte infusion (DLI)-based for 40/103 (39%) patients. Results: The 1-, 2-, and 5-year OS of patients treated with DLI (n = 40) was 67%, 34%, and 34%, respectively, for those treated preventively (n = 20) and 43%, 20%, and 20%, respectively, for those treated in overt relapse (n = 20) (p < 0.01). The 1-, 2-, and 5-year OS of patients treated without DLI (n = 63) was 54%, 40%, and 26%, respectively, for those treated preventively (n = 14) and 17%, 5%, and 0%, respectively, for those treated in overt relapse (n = 49) (p < 0.01). Conclusions: Relapse treatment with a pre-emptive strategy was associated with improved outcomes, particularly when DLI was employed.

The primary factors associated with a higher risk of post-transplant relapse encompass the absence of complete remission (CR) at the time of allo-SCT, high-risk cytogenetic characteristics, T-cell depletion, and the use of RIC regimens [11][12][13][14].The current challenge lies in identifying the optimal strategy to manage the post-transplant phase in order to reduce disease recurrence risk or prevent it [15].Interventions can be employed in three distinct settings: the maintenance phase, when there is no evidence of relapse; the pre-emptive phase in the presence of detectable minimal residual disease (MRD) or mixed chimerism to prevent overt relapse; or the therapeutic phase, when overt disease relapse has occurred [15][16][17].Cellular immunotherapy, such as donor lymphocyte infusion (DLI), a second transplant, as well as the use of hypomethylating agents (HMAs) and innovative targeted therapies are the available therapeutic post-transplant options [18][19][20][21][22][23][24][25].The selection of one treatment over another frequently relies on the patient's clinical factors, including performance status and the presence of GVHD, as well as the type of relapse (early or overt).Due to the multitude of variables involved, a precise algorithm for the timing and selection of these treatments has not yet been defined [8].
In this scenario, we analyzed a cohort of 134 AML and MDS adult patients consecutively allotransplanted between 1 January 2015, and 31 December 2021 at Hospital Saint-Antoine AP-HP and Spedali Civili di Brescia and who subsequently relapsed.The aims of this study were to evaluate patient survival after allo-SCT relapse, to describe the type of relapse, and to identify the best therapeutic strategy to improve the outcome post-allo-SCT relapse.

Study Design and Patients
From January 2015 to December 2021, a total of 553 AML/MDS patients were consecutively submitted for allo-SCT in the Bone Marrow Transplant Unit of Spedali Civili di Brescia (Italy) and Saint-Antoine University Hospital, Paris (France).Of these, 134 relapsed after allo-SCT (24.2%) and 103 (18.6%) were subsequently treated.Patients were categorized according to the type of relapse into two groups: the overt relapse (69/103 patients-67%) and the early relapse, which included patients with molecular/cytogenetic relapse and/or with mixed chimerism (34/103 patients-33%).Furthermore, within these two groups, we further divided patients based on whether they had received DLI (40/103-38.8%) or not (63/103-61.2%).All patients provided informed consent for data registration in the PROMISE database, in which clinical and biological data are collected.Supplementary data were extracted through a comprehensive review of patients' clinical charts, encompassing both the transplant phase and the subsequent follow-up period.The study was conducted in compliance with current National and European legislation on clinical trials, in accordance with the Declaration of Helsinki and the principles of good clinical practice.

Definitions
Overt relapse was defined by the recurrence of blasts in peripheral blood (PB) or bone marrow (BM) infiltration by more than 5% of blasts.Early relapse was characterized by molecular or cytogenetic recurrence and/or mixed chimerism.Pre-emptive treatment was defined as therapy administration in cases of early relapse [26].DLI was defined as the transfusion of unstimulated lymphocyte concentrates collected from the original stem cell donor as buffy coat preparations or as a transfusion of unmanipulated mobilized stem cells.
At Spedali Civili di Brescia, the median dose of the first DLI was 1 × 10 6 CD3+ cells/Kg, and in case of multiple infusions, an escalating schedule was chosen (5 × 10 6 CD3+ cells/Kg for the second infusion, 10 × 10 6 CD3+ cells/Kg for the third and 50 × 10 6 CD3+ cells/Kg for the fourth).Lymphocyte doses in the case of haploidentical transplantation were reduced by 1 Log.
At Saint-Antoine University Hospital, the DLI doses were adapted depending on the setting (prophylaxis, pre-emptive, overt disease) and donor type (sibling, matched unrelated donor (MUD), mismatched unrelated donor (MMUD) or haploidentical).Briefly, focusing on the sibling and MUD 10/10 setting, the first dose of prophylactic DLI was 1 × 10 6 CD3+/Kg, increasing by a half log each for the subsequent two doses.In the same setting of prophylaxis, considering MMUD and haploidentical donors, each dose level was reduced by a half log and by 1 Log, respectively.Moving from the prophylactic setting to the pre-emptive and overt disease setting, each dose level for each donor type was increased by a half log.
The chimerism assessment on Italian patients in the Spedali Civili di Brescia laboratory was evaluated on BM CD34+cells by RT-qPCR (reverse transcription-quantitative polymerase chain reaction) short tandem repeat analysis [27,28].The chimerism assessment on French patients in Saint-Antoine University Hospital was evaluated on PB CD33+ and CD3+ cells by RTqPCR [8,9].Mixed chimerism was defined as failure to achieve >97.5% of donor cells, following data from our previous paper, which suggested this cut-off was able to significantly predict relapse [10].Molecular monitoring of MRD was performed on common target genes (WT1, NPM1, FLT3-ITD) on PB or BM by RT-qPCR if these genes were detected at diagnosis.Complete remission (CR) was defined as the presence of <5% blasts in the BM and no circulating blasts in the PB [1].Molecular complete remission (CRmol) was defined as MRD negativity [1].Moreover, for analysis purposes, patients who re-obtained full donor chimerism were grouped together with those who obtained Crmol.

Statistical Analysis
Categorical data were presented as numbers and percentages, continuous data as median and range, respectively.The chi-squared test was used to test for differences among subgroups.The OS, defined as the interval from the date of post-allo-SCT relapse to death or last follow-up, was estimated using the Kaplan-Meier method [29].The log-rank and Gray's tests were employed to verify differences among the different groups.Univariate and multivariate analyses were performed using a Cox regression model [29].In the univariate analysis, variables considered as possible prognostic factors were disease status at allo-SCT (CR vs. no CR), post-relapse therapy (yes vs. no), conditioning intensity (myeloablative vs. RIC), DLI administration (yes vs. no), acute GVHD (aGVHD) and chronic GVHD (cGVHD) after allo-SCT, donor type (sibling vs. MUD plus haploidentical donor), time of allo-SCT relapse > 1 year, and the type of relapse (overt vs. early).Multivariate analysis included all variables found to be significant at the level of p < 0.05.Statistical analyses were performed with EZR (version 4.2.2) [30].

Patient and Transplant Characteristics
After allo-SCT relapse, 103 (77%) patients received therapy, and 31 (23%) did not due to their precarious clinical conditions.Table 1 shows the most important clinical and transplant characteristics of patients treated after allo-SCT relapse.The characteristics are listed according to the type of relapse, distinguishing between overt (n = 69) and early relapse (n = 34).In the overt relapse group, the median age at transplant was 60.1 years (20-74), and 39 patients (56.5%) were male.84.1% of patients were diagnosed with AML, while 15.9% were diagnosed with MDS.50.7% of patients received allo-SCT in CR, and 55.1% were in the first RC.PBSC was used in 94.2% of the cases, and the donor was sibling in 27.5% of the transplants, MUD/MMUD in 34.8% and haploidentical in 36.2%.The regime of conditioning was MAC in 53.6% of patients and RIC in the remaining 46.4%.GVHD prophylaxis consisted of cyclosporine (CsA) plus methotrexate (MTX) in 3 patients (4.3%), CsA plus MTX plus antithymocyte globulin (ATG) in 29 patients (21.7%),CsA plus mycofenolato mofetile (MMF) plus ATG in 16 patients (23.2%), post-transplant Cyclophosphamide PTCy based prophylaxis in 25 patients (36.2%), and other platforms in the remaining 10 patients.From the allo-SCT to the last follow-up, acute-GVHD (aGVHD) occurred in 34 (49.3)patients and chronic-GVHD (cGVHD) in 10 (14.5).The median time to relapse was 12 months (0.8-60.5).
In the early relapse group, the median age at transplant was 60.0 years (28.9-67.5),21% of patients (61.8%) were male, and 31 (91.2%) were diagnosed with AML.50% of patients received allo-SCT in CR, and 61.8% were in the first RC.PBSC was used in 91.2% of the cases, and the donor was sibling in 38.2% of the transplants, MUD/MMUD in 44.1% and haploidentical in 17.6%.The regime of conditioning was MAC in 55.9% of patients and RIC in the remaining 44.1%.GVHD prophylaxis consisted of CsA plus MTX in 3 patients (8.8%), CsA plus MTX plus ATG in 14 patients (41.2%),CsA plus MMF plus ATG in 8 patients (23.5%), post-transplant Cyclophosphamide PTCy based prophylaxis in 5 patients (14.7%), and other platforms in the remaining 4 patients.aGVHD was observed in 13 (38.2) patients and cGVHD in 11 (32.4) from the allo-SCT to the last follow-up.The median time to relapse was 12 months (1.0-56.3).

Discussion
Relapse following allo-SCT still represents the greatest obstacle against AML/MDS cure.As early as 2007, Schmid et al. [31] explored the role of DLI in the treatment of AML in overt hematological relapse.The OS at 2 years was 21% for patients receiving DLI and 9% for patients not receiving DLI.Among DLI recipients, factors such as lower tumor burden at relapse and remission at the time of DLI were predictive for survival in multivariate analysis.These findings prompted clinicians to identify the relapse as early as possible to adopt a pre-emptive treatment strategy [32].The combination of MRD monitoring with lineage-specific molecular chimerism analysis appears to be the most sensitive way of detecting disease recurrence following allo-HCT [33].Thus, in a recent work, the OS rate at 5 years after pre-emptive DLI for MRD/mixed chimerism was between 51% and 68% among responders and 37% among non-responders [34].
In this series of 134 AML/MDS patients relapsing after allo-SCT, the 2-year OS rate after allo-SCT relapse was estimated at 18%, which falls within the range reported in the existing literature (Figure 2a) [19,31,35].In a recent European Society for Blood and Marrow Transplantation (EBMT) registry study on 8162 adult patients with AML who relapsed between 2000 and 2018 after allo-HCT, Bazabachi et al. [9] confirmed a dismal 2-year OS of around 17% for the entire cohort and observed a steady increase in 2-year survival from 16% to 26% among patients aged ≤ 50 years at relapse, likely reflecting, among other factors, the efficacy of post-transplant salvage including second allo-HCT.As expected, in our study, most (89%) deaths were attributed to the progression of the disease, highlighting the difficulty of relapse management.Nevertheless, treating patients with cellular therapy and/or innovative drugs after allo-SCT relapse results in better outcomes compared to palliative care, as shown in Figure 2b, and this prompts us to invest resources in identifying the best post-transplant salvage strategy.
To better understand how and when to act to improve survival after disease recurrence, we divided the patients who received therapy into two groups, based on type of relapse: overt and early, the latter including patients with recurrence of molecular/cytogenetic disease and/or with mixed chimerism (<97.5% donor cells).These two cohorts had almost completely overlapping characteristics.In the overt relapse group, haploidentical donors were more frequent, resulting in a higher number of patients who received PTCy as GVHD prophylaxis.While the difference in aGVHD was not statistically significant between the two groups, more patients treated in early relapse had cGVHD (p = 0.04) due to the higher number of patients receiving DLI in this context [34].The better OS of patients treated in pre-emptive setting confirms the immunological efficacy of DLI in treating posttransplant leukemia recurrence, and also the central role of MRD monitoring after allo-SCT combining the early detection of molecular specific disease markers with post-transplant chimerism modifications.
The choice of therapy, including DLI or not, was based on the local policies adopted by the two centers during the last 6 years, largely influenced by the type of recurrence since DLIs are conventionally more frequently used when the recurrence is molecular or the chimerism is mixed [34,36] and in the absence of ongoing GVHD.Patients treated with DLIbased regimens had a significantly better OS compared to those treated with other therapies.The difference in OS between patients who received DLI only and patients treated with DLI in combination with other therapies reflects the early phase of relapse in the former group; notably, eight out of nine patients who received only DLI had early relapse.
When the OS is analyzed based on DLI administration (yes vs. no) and the phase of treatment administration (early vs. overt relapse), incorporating DLI into treatment regimens for overt relapse significantly improves the survival approach of patients treated in a pre-emptive setting.The use of new drugs such as epigenetic modulators, venetoclax, or FLT3 inhibitors likely enables a greater number of patients to reduce the leukemic burden and achieve remission, which can be further consolidated with subsequent DLI infusions.Furthermore, these data confirm the potentially curative effectiveness of graft-versusleukemia even against advanced disease relapses [31,[37][38][39].As expected [40,41], the best outcomes were achieved when the relapse was promptly detected, and pre-emptive therapy was started, particularly if DLI was administered.In this latter case, it is worth noting that patients treated in early relapse with DLI reached a plateau in OS, which suggests that they may be cured.
Some limitations of our study must be considered.Firstly, patients were retrospectively evaluated; secondly, our results were derived from the analysis of subgroups that comprised relatively small numbers.This demands caution in drawing definitive conclusions and suggests that further studies on a larger number of patients are needed.Finally, MRD and chimerism were measured locally using different methods.Nevertheless, these results are in line with other data and suggest that even in the absence of randomized trials, the issue of MRD monitoring and pre-emptive treatment of relapse in the real world can be considered a mainstay of Good Clinical Practice.
In summary, our data confirm the poor outcome of AML/MDS patients who relapse after allo-SCT and show the real survival benefit obtained from a pre-emptive treatment strategy [26].Which drugs or drug combinations are more effective has not yet been established and will probably be the focus of future studies.Moreover, the detection of MRD with highly specific and sensitive methods (e.g., digital PCR) [42,43] should be implemented in the future in order to progressively increase the number of patients who are treated pre-emptively with DLI.

Conclusions
The number of therapeutic opportunities to reduce the risk of disease relapse after allo-SCT for AML and MDS has increased in recent years, but the improvement in outcome following relapse after transplant is still an unmet clinical need.Integrating MRD detection after allo-SCT with chimerism monitoring and adopting an immunological pre-emptive intervention with DLI either alone or in combination might be a successful strategy [4].The future priority must be the design of randomized trials, exploring the role of new targeted drugs and immunomodulating agents such as DLI in the different scenarios of the post-transplant follow-up, starting from the prophylaxis of disease relapse, and moving to the treatment of MRD (pre-emptive therapy) before overt relapse [26].

Figure 1 .
Figure 1.Cumulative incidence of any grade acute and chronic GVHD after DLI-based therapy.

Figure 1 .
Figure 1.Cumulative incidence of any grade acute and chronic GVHD after DLI-based therapy.

Figure 1 .
Figure 1.Cumulative incidence of any grade acute and chronic GVHD after DLI-based therapy.

Figure 4 .
Figure 4. (a) Overall survival (OS) of the 9 patients treated after allo-SCT relapse with DLI only (black line) compared with OS of 31 patients treated with DLI plus other therapies (red line) and OS of 63 patients treated without DLI (green line).(b) OS of the patients treated with DLI-based regimens in early relapse (green line) versus overt relapse (blue line) versus OS of the patients treated without DLI in early relapse (black line) versus overt relapse (red line).

Figure 4 .
Figure 4. (a) Overall survival (OS) of the 9 patients treated after allo-SCT relapse with DLI only (black line) compared with OS of 31 patients treated with DLI plus other therapies (red line) and OS of 63 patients treated without DLI (green line).(b) OS of the patients treated with DLI-based regimens in early relapse (green line) versus overt relapse (blue line) versus OS of the patients treated without DLI in early relapse (black line) versus overt relapse (red line).

Table 1 .
Clinical and transplant characteristics of the 103 patients treated after allogeneic stem cell transplant relapse.

Table 2 .
Post-relapse therapy of the 103 patients included in the study according to type of relapse.

Table 3 .
Response to therapy of 103 patients treated after allo-SCT relapse according to type of relapse.