Introduction

Allogeneic hematopoietic cell transplantation (allo- HCT) is a potentially curative therapy for acute leukemia¹. However, patients with acute leukemia who relapse after allo-HCT show poor prognosis with a median survival of 34 months². A second allo-HCT results in a longterm survival rate of only 10-35% with a higher treatment- related mortality. Currently, no standard treatment approach is available for relapsed acute leukemia^3,4.

Induction of graft-versus-leukemia (GVL) effects with donor lymphocyte infusions (DLIs) is an attractive option for patients with relapsed hematological malignancies. However, GVL efficacy depends on disease subtype and tumor burden at the time of DLI^3,5. Schmid et al. demonstrated an overall survival (OS) benefit of DLI in patients with acute myelogenous leukemia (AML) who relapsed after allo-HCT (20% ±3% vs. 9% ±2%; P<0.001) ⁶. Several different strategies have been explored to improve patient outcomes, such as dose-escalation of DLIs, addition of immunosuppressive agents to prevent graft-versus- host disease (GVHD), and modified DLI treatment with granulocyte colony-stimulating factor (G-CSF) ^7-10. A combination of chemotherapy and DLI showed promising results¹¹. Here, we investigated the effect of cytoreduction with a high dose of cytarabine followed by DLI. A pilot study performed at our institution reported a beneficial effect of cytarabine combined with G-CSF-primed DLIs using cryopreserved cells on patients with hematological malignancies who relapsed after allo-HCT¹². This strategy was cost-effective and convenient for donors. In the current study, we aimed to determine the effectiveness of cytarabine in combination with G-CSF-primed DLI in treatment of patients with acute leukemia who relapsed after allo-HCT.

Materials and Methods

Data Collection

We conducted a retrospective review of the medical records of 255 patients who received allo-HCT for AML, myelodysplastic syndrome (MDS), or acute lymphoblastic leukemia (ALL) between December 1998 and August 2013 at the Department of Hematology/Oncology, Kyungpook National University Hospital (KNUH). Clinical and laboratory data were collected from electronic medical records according to protocol approved by the KNUH institutional review board.

Definitions

The risk status at transplantation was determined based on previously published classification schemes¹³. Poorrisk cytogenetics were classified according to the revised Medical Research Council classification system for AML and the International Prognostic Scoring System for MDS^14,15. Poor-risk cytogenetics for ALL were defined as MLL rearrangement, BCR/ABL1 translocation, hypoploidy, or complex karyotype. Graft failure was defined as the lack of myeloid engraftment in patients surviving in remission for at least 28 days after transplantation. The Keystone staging system was used to score acute GVHD (aGVHD) and chronic GVHD (cGVHD) ^16,17. Relapse was defined as the reappearance of leukemic cells in the peripheral blood, bone marrow, or extramedullary lesions after allo-HCT.

A novel composite end-point of refined GVHD-free/ relapse-free survival (GRFS), where events included grade ?-? aGVHD, systemic therapy requiring cGVHD, relapse, or death, was also¹⁸. OS was calculated from the date of the first allo-HCT to the date of death, or to the last follow-up. Relapse-free survival (RFS) was calculated from the date of the first allo-HCT to the date of disease recurrence or to the date of death. Post-relapse survival (PRS) was defined as the time from relapse posttransplantation to death or to the last follow-up¹⁹.

Transplantation procedures

Preparative regimens for allogeneic peripheral blood stem cell transplantation (PBSCT) included busulfan (Bu, 4 mg/kg PO or 0.8 mg/kg IV for 4 days) and cytoxan (Cy, 60 mg/kg for 2 days) administered to 100 patients; Bu (3.2 mg/kg for 2-4 days) and fludarabine (Flu, 30 mg/m² for 6 days) administered to 135 patients; and total body irradiation and Cy (60 mg/kg for 2 days) administered to 20 patients. PBSCs were mobilized with 10μg/kg per day G-CSF[filgrastim (Leukokine®; CJ, Co., Korea) or lenograstim (Neutrogin®; Chugai Co. Ltd, Tokyo, Japan) ]alone (n=183, 71.8%) or in combination with a concurrent regimen of 5μg/kg per day G-CSF and 5μg/kg per day granulocyte macrophage colony-stimulating factor (GM-CSF) (n=72, 28.2%) from the donor. Administration of G-CSF with or without GM-CSF was continued, and apheresis was repeated every morning until the targeted cell dose (6×10⁶ CD34+ cells/kg) was reached. GVHD prophylaxis consisted of methotrexate ( MTX) and cyclosporine A (CyA) or MTX and tacrolimus (Tac).

Collection and infusion of donor lymphocytes

Collecting the targeted amount of PBSCs (more than 6×10⁶ CD34+ cells/kg) allowed us to cryopreserve some PBSCs, including several CD3+ cells at the time of harvest for transplantation. The extra cells were cryopreserved with dimethyl sulfoxide in a nitrogen tank. DLI with cryopreserved cells was available only for patients who were able to receive more than 6×10⁶ CD34+ cells/kg from donors. Other patients would receive salvage chemotherapy or second transplantation when they relapse.

For those patients who relapsed after allo-HCT, DLI was promptly performed using the cryopreserved cells. The CD3+ cell-count was determined by flow cytometry and used to calculate the DLI dose. Before DLI, immunosuppressive agents were discontinued and patients received pre-DLI chemotherapy with high-dose cytarabine ( 1 g/m²; twice a day on days 1, 3, and 5). No patient received prophylactic immunosuppressive therapy after DLI.

The chimerism status, which was assessed by the number of tandem repeats or short tandem repeats, was compared before and after DLI. All patients underwent a bone marrow examination within 60 days after DLI or sooner (if clinically indicated) for the assessment of the response.

Statistical Analysis

Categorical data were analyzed using a chi-square test. Survival analysis was conducted using the Kaplan-Meier method, and groups were compared using a log-rank test. The cumulative incidence of GVHD was calculated using the Gray method considering treatment-related mortality and relapse as competing risks. The Cox proportional regression model was used to analyze potential risk factors affecting survival. Statistical analyses were per- formed using the SPSS software version 18 (SPSS Inc., Chicago, IL, USA) and EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan) ²⁰.

Results

Patient and transplant characteristics

A total of 255 patients were analyzed. The median infused cell doses were as follows: mononuclear cells, 7.94×10⁸ cells/kg (range: 0.36-25.12) ; CD34+ cells, 5.13×10⁶ cells/kg (range: 0.46-20.6) ; and CD3+ cells, 2.82×10⁸ cells/kg (range: 0.05-10.0). Patients were reclassified into two groups according to the targeted CD34+ cell dose (6×10⁶ cells/kg) according to the KNUH protocol. The lower CD34+ group (n=165; 64.7%) included patients who underwent allo-HCT with a CD34+ cell dose of lower than 6×10⁶ cells/kg, and the higher CD34+ group (n=90; 35.3%) included patients who underwent allo-HCT with a CD34+ cell dose of at least 6×10⁶ cells/kg. Patient characteristics are summarized in Table 1. No statistically significant differences in the transplantation outcomes, such as the incidence of aGVHD, cGVHD, and relapse rate, were found between the two groups (Table 2).

Table 1. Patient characteristics according to CD34+ cell dose

ECOG PS, Eastern Cooperative Oncology Group performance status; AML, acute myelogenous leukemia; MDS, myelodysplastic syndrome; ALL, acute lymphoblastic leukemia; CR, complete remission; CMV, cytomegalovirus; G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte macrophage colony-stimulating factor; GVHD, graft-versus-host disease; CsA, cyclosporine;MTX, methotrexate; TCD, T-cell depletion; ATG, anti-thymoglobulin.

Table 2. Transplantation outcomes according to CD34+ cell dose

Others: 3 patients died of cardiac arrest, 2 patients died of brain hemorrhage, 1 patient died of pulmonary hemorrhage, 1 patient died of malnutrition, 1 patient died of acute renal failure. MNC, mononuclear cell; aGVHD, acute graft-versus-host disease; cGVHD, chronic graft-versus-host disease; VOD, veno-occlusive disease.

Impact of CD34+ cell dose on GRFS

The median follow-up duration was 18.1 months with a range of 0.2 to 209.7 months. The 1-year OS, RFS, and non-relapse mortality (NRM) were 55.3% ±3.1%, 66.0% ±3.2%, and 28.2% ±0.3%, respectively. The cumulative incidences of aGVHD and cGVHD were 40.7% ±0.3% and 41.6% ±0.3%, respectively. The unadjusted Kaplan-Meier estimate of 1-year GRFS was 32.9% ±3.1%. No significant difference was found in OS, RFS, NRM, or GRFS between the two groups classified according to the CD34+ cell dose (Figure 1). Moreover, there was no significant correlation between the number of infused CD3+ and CD34+ cells (Spearman correlation coefficient, P=0.307). However, a trend of more CD3+ cells (more than 3.1×10⁸ cells/kg) was noted in the higher CD34+ group (P=0.001; Table 2). In the univariate analysis, patients transplanted with higher CD34+ and CD3+ cell doses did not show an improved GRFS (Table 3, P=0.623 and P=0.158, respectively). The risk status at transplantation was an independent factor associated with worse GRFS (hazard ratio (HR) =1.782, 95% confidence interval (CI) : 1.267-2.509, P=0.001; Table 3).

Figure 1. Survival curves according to infused CD34+ cell dose

Patients were divided into lower or higher CD34+ dose group based on cut-off dose(6×10⁶/kg)of CD 34+ cells(. A)OS: 1-year OS rates were 53.3% ±3.9% and 58.9% ±5.2% for the lower and higher groups, respectively(P=0.696)(. B)RFS: 1-year RFS were 66.7% ±4.0% and 63.2% ±5.5% for the lower and higher groups, respectively(P=0.332)(. C) NRM: 1-year NRM were 29.5% ±3.8% and 25.6% ±4.7% for the lower and higher groups, respectively(P=0.939)(. D)GRFS: 1-year GRFS were 35.3%±4.0% and 28.4% ±5.1% for the lower and higher groups, respectively(P=0.245).

Table 3. Factors affecting GRFS

GRFS, graft-versus-host disease-free, relapse-free survival; HCT, hematopoietic cell transplantation; TCD, T-cell depletion; RIC, reduced-intensity conditioning; NMA, non-myeloablative; MAC, myeloablative conditioning; MUD, matched unrelated donor; MSD, matched sibling donor; FD, female donor; MR, male recipient; CMV, cytomegalovirus.

Post-Relapse Survival

Among the 255 patients, 93 (36.4%) relapsed after allo-HCT. The median time from allo-HCT to relapse was 4.6 months (range: 1.5-59.1). After relapse, 45 patients (48.4%) were treated with salvage chemotherapy with a regimen based on fludarabine or mitoxantrone (FLAG or MEC), 9 (9.7%) with a second allo-HCT because of unavailability of reserved cells for DLI, and 39 (41.9%) with G-CSF-primed DLI (4 patients received DLI only because of their refusal on the use of cytarabine). Thereafter, 13 patients (30.0%) achieved DLIinduced complete remission, 24 progressed, and 2 were not evaluable for response. DLI-induced aGVHD was observed in 24 patients (61.5%) with a median of 20 days after DLI (range: 3-98 days) ; ten patients with grade ?, six with grade ?, five with grade ?, and three with grade ?. As shown in Table 4, univariate analysis revealed that poor-risk cytogenetics (HR=2.512, P= 0.015), risk status at transplantation (HR=4.406, P< 0.001), myeloablative conditioning regimen (HR=0.567, P=0.007), cGVHD (HR=0.525, P=0.006), and longer post-transplantation remission duration (HR=0.297, P< 0.001) were significantly associated with PRS. A longer post-transplantation remission duration was the only independent factor correlated with PRS (HR=0.297, 95% CI: 0.193-0.457, P<0.001).

Table 4. Factors affecting Post-Relapse Survival(PRS)*

Post-relapse survival(PRS)was defined as the time from relapse post-transplantation to death or last follow-up. **Patients were reclassified into two groups according to the targeted CD34+ cell doses(6×10⁶/kg)by the KNUH protocol. ***The median CD3+ cell dose was 2.82×10⁷/kg(range: 0.05-10.1). ****Post-transplantation remission duration was divided by the median RFS 4.6 months(range:1.5-59.1). ECOG PS, Eastern Cooperative Oncology Group performance status; TCD, T-cell depletion; MUD, matched unrelated donor; MSD, matched sibling donor; MMSD, mis-matched sibling donor; FD, female donor; MR, male recipient; CMV, cytomegalovirus; aGVHD, acute graft-versus-host disease; cGVHD, chronic graft-versus-host disease; PLT, platelet; DLI, donor lymphocyte infusion.

G-CSF-primed DLI effect on PRS

Among the 39 patients (41.9%) who received DLIs, 34 received one infusion and 5 received two infusions. The median amount of CD3+ cells was 2.82×10⁷ cells/ kg (range: 0.05-10.1). The patient and transplant characteristics according to the post-transplantation remission duration are described in Table 5. Patients were classified into early and late relapse groups based on post-transplantation remission duration relative to the median RFS of 4.6 months (range: 1.5-59.1). For patients with early relapse (remission duration<4.6 months), one-year PRS rates were 7.7% ±5.3% and 5.3% ±4.3% in the DLI and non-DLI groups, respectively (Figure 2A; P=0.667). For patients with late relapse (remission duration?4.6 months), one-year PRS rates were 40.0% ±7.4% and 17.9% ±7.2% in the DLI and non-DLI groups, respectively ( Figure 2B; P=0.039). The number of second allo-HCT cases was small (n=9), they were, therefore, excluded from comparison analysis.

Table 5. Outcomes of DLI treatment according to post-transplantation remission duration

Post-transplantation remission duration was divided by the median RFS of 4.6 months

Figure 2. Post-Relapse Survival(PRS)according to the use of DLI

Early relapse and late relapse groups were determined based on median post-transplantation remission duration(. A)In the early relapse group, 1-year PRS rates were 7.7% ±5.3% and 5.3% ±4.3% for DLI and non-DLI groups, respectively(P=0.667)(. B)In the late relapse group, 1-year PRS rates were 40.0% ±7.4% and 17.9% ±7.2% for DLI and non-DLI groups, respectively(P=0.039).

Discussion

The current study investigated the efficacy of cytarabine- based chemotherapy in combination with G-CSFprimed DLI in patients with acute leukemia who relapsed after allo-HCT. G-CSF-primed DLI treatment after allo- HCT with a limited CD34+ cell dose (lower than 6×10⁶ cells/kg) constituted a feasible and effective option in terms of GRFS, donor convenience, and cost. Moreover, this treatment option may replace a second HCT for laterelapse patients. Although stem cell dose has already been explored in relation to the incidence of GVHD, relapse, and survival, no consensus has been reached^21-23.

Regarding CD 34+ cell dose for allogeneic PBSCT, higher CD34+ dose (higher than 8×10⁶ cells/kg) resulted in better neutrophil and platelet recovery, and was associated with development of chronic GVHD, but was not correlated with improved survival. Optimal CD34+ dose seemed to be dependent on donor type and stem cell source²⁴.

Preliminary results from our institution demonstrated that CD34+ cell transplantation with a dose of higher than 6×10⁶ cells/kg did not improve refined GRFS (median survival 5.5 months vs. 6 months, P=0.245; Figure 1D). Moreover, a higher CD34+ cell dose did not increase the neutrophil or platelet engraftment rate. As the current study found no correlation between the CD3+ and CD34+ cell numbers in the harvested cells (Spearman correlation coefficient, P=0.307), it is planned to limit the CD34+ cell dose (6×10⁶ CD34+ cells/kg) for transplantation and cryopreserve the rest of the harvested cells for relapse or prophylactic use.

Cryopreservation for the purpose of future DLI was not performed in patient group who received a dose of lower than 6×10⁶ CD34+ cells/kg in the initial allo-HCT.

This retrospective study has several limitations, including the heterogeneity of patients and transplant characteristics. Furthermore, DLI treatment has a minimal effect in the case of a rapidly advancing disease, as evidenced in patients with early relapse who experienced no benefits from DLIs. However, patients with longer post-transplantation remission duration showed better PRS in the DLI group (Figure 2B; 1-year OS: 46.7% ±12.9% vs. 21.7% ±8.9%, P=0.039).

A second allo-HCT is regarded as the optimal option for patients who relapse after the first transplantation. Yet, it is only available for a limited number of patients due to concerns of high mortality and unavailability of donors. Thus, for the late relapse group, DLI treatment may replace second HCT. A faster recovery can also be expected in the case of chemotherapy followed by G-CSF-primed DLI treatment with a sufficient number of CD34+ cells. Regarding GRFS, allo-HCT with a limited CD34+ cell dose (lower than 6×10⁶ cells/kg) is not inferior to allo-HCT with a higher CD34+ cell dose. Moreover, the surplus cells from the harvest can be cryopreserved at the time of the first transplantation. Then, DLI treatment using these cryopreserved cells can be promptly performed at the time of relapse without a need for a new harvest. From the perspective of donor convenience and cost-effectiveness, this strategy constitutes an attractive option for patients with dismal prognosis after post-transplantation relapse.

Well-designed prospective clinical trials are required to answer such DLI-related questions as to when, how, and to whom. Previous studies have shown multiple biological effects of G-CSF on peripheral blood stem cells, including the ability to polarize T cells from Th1 to Th2, promotion of regulatory T cells, and tolerogenic dendritic cell differentiation^25,26. In addition, this study indicated that G-CSF-primed DLI, rather than unstimulated DLI, included more CD34+ cells and led to early recovery. Moreover, interestingly, low mortality was associated with DLI-induced GVHD, and most of the mortality resulted from disease relapse or refractory disease rather than GVHD.

According to a study from Japan, in the case of nonprimed DLI²⁷, complete remission (CR) rate was 38% and the probability of remaining in CR at 3 years was 7% in AML. Acute GVHD (grade 2 or higher) developed in 31 out of 89 (35%) patients with HLA-identical related donors and was fatal for seven (8%). The incidence of aGVHD (cumulative incidence of aGVHD; 40%) was similar to that in the current report. Chronic GVHD developed in 24 of 73 (33%) patients who received DLI from HLA-identical relatives.

Recently, Claiborne et al.²⁸ reported that treatment of relapsed AML/MDS (n=28) after allo-HCT with acombination of azacytidine and DLI resulted in a two-year OS rate of 35%, and noted a trend towards higher absolute CD4+ cell count in the patient group achieving remission. Hypomethylating agents may induce synergistic effect by promoting cytotoxic effects against leukemic cells when combined with DLI²⁸. Schroeder et al. also reported complete and partial remission rates of 27% and 6%, respectively, which correspond to an overall response rate of 33%, following the treatment of relapsed AML/ MDS after allo-HCT with a combination of azacytidine and DLI. Two-year OS rate was higher in MDS patients and correlated with disease burden in patients with AML. Overall incidence of aGVHD after treatment with Aza and DLI was 23% among all patients included in the study (n=154) and 31% among patients who had received DLI (n=105). The authors concluded that the combination of azacytidine and DLI was an effective and well-tolerated treatment option for patients with relapse after allo-HCT, in particular for those with MDS or AML with low disease burden²⁹. Another group, which tested a combination of decitabine and DLI, reported that the effectiveness of this regimen was not restricted to patients with low leukemic burden³⁰.

In conclusion, our results indicate that G-CSF-primed DLI treatment after allo-HCT with a limited CD34+ cell dose (lower than 6×10⁶ cells/kg) is a feasible and effective option in terms of GRFS, donor convenience, and cost. Therefore, it has potential to replace the second HCT in treatment of late relapse patients.

Author's Contribution

As the first author, Y. J. L. collected and interpreted the data, drafted the article, and critically revised the important intellectual content; Y. J. L., D. W. B., H. J. J., J. H. M. and S. K. S. performed the treatment, supplied the acquisition of data, and revised the manuscript; S. K. S. provided the conception and design of the study, critically revised the article for important intellectual content, and gave the final approval of the version to be submitted.

Conflict of Interest

The authors declar no conflict of interest. Disclosure forms provided by the authors are available here.

Financial Support

This research was supported by Kyungpook National University Research Fund, 2018-2020.