Peripheral blood used as a source of stem cells for transplantation (PBSCT) is known to exert stronger immune-mediated effects compared with BM (BMT). We decided to retrospectively analyze the impact of stem cell source on the OS of CML patients who relapsed after either matched related donor PBSCT (N=168) or BMT (N=216) and were treated with donor lymphocyte infusions (DLI). Univariate analysis revealed a lower probability of OS after DLI in patients relapsing after PBSCT vs BMT (66% vs 79% at 5 years, P=0.013). However, a multivariate Cox analysis did not reveal any significant impact of PBSCT as a risk factor for decreased OS for patients transplanted in first chronic phase (CP1; hazard ratio (HR) 1.036, 95% confidence interval (CI) 0.619–1.734). A statistical interaction term suggested that the impact of stem cell source on OS after DLI was different for those transplanted in advanced phases (negative impact of previous PBSCT—HR 2.176, 95% CI 0.930–5.091). In summary, the stem cell source does not affect the OS of CML patients who underwent PBSCT in CP1, relapsed and were treated with DLI. However, when the patients were transplanted in advanced phases, previous PBSCT seems to negatively affect OS after DLI compared with BMT.
The strategy of alloSCT in CML has evolved over the last two decades both in terms of the practical approach to treatment protocol and indications for this procedure.1 It is no longer considered a standard of care for the majority of otherwise SCT eligible patients, but it is currently applied to selected subgroups for example, with advanced phases of CML, with high-risk disease or disease in chronic phase resistant to tyrosine kinase inhibitors (TKIs). The predominating source of hematopoietic stem cells has also switched from BM to PBSC. As a result, in 71% of alloSCT performed in 2009 in Europe, the PBSC source was used.2 This switch to PBSC transplantation (PBSCT) has been influenced by its practical advantages, which include higher efficacy of cell collection and more rapid neutrophil and platelet recovery after transplantation.3 Moreover, the source of hematopoietic stem cells may affect the outcomes of alloSCT.3, 4 Peripheral blood (PB), most likely due to a higher content and different profile of T lymphocytes, exerts more pronounced immune effects on the stem cell recipient than BM. The clinical consequences of this are a stronger GvL effect on the one hand but also an increased risk of GvHD on the other, and these differences are especially evident in CML.
CML is a good example of a disease that is highly susceptible to allogeneic immune effects.1 However, the efficacy of this treatment is largely decreased in the case of transplantation of patients who have advanced-phase CML. Interestingly, in those patients, the OS and leukemia-free survival rates after PBSCT have been shown to be higher compared with BMT.3, 4, 5 Schmitz et al.4 reported 25% and 33% 6-year probabilities of OS for advanced CML after BM and PBSC transplantation, respectively, while the corresponding probabilities for early CML were 64% and 45%. Therefore, for advanced phase CML, the increased GvL effect seems to outweigh the risk of GvHD-related mortality. However, in the first chronic phase (CP1), the benefit of immune reactivity of PBSCT seems to be decreased due to the higher rate of late transplant-related mortality caused by cGvHD. Based on this, it was suggested that for patients transplanted in CP1, BM should be the preferred choice of stem cells and PB should be used for those transplanted in advanced phase.1
The exploitation of the GvL effect by the use of donor lymphocyte infusions (DLI) is currently a standard approach to treating CML relapse after alloSCT. In CML, this treatment is very efficacious and has been shown to induce remission in 60–87% of patients of which about 90% are believed to be cured.6, 7, 8, 9 However, only 58% of patients are estimated to be alive after 5 years following DLI, and the main causes of mortality are CML relapse and complications of this procedure.6 A number of factors were shown to be related with improved survival after DLI: long interval between SCT and relapse, chronic CML phase at relapse, molecular or cytogenetic grade of relapse, low initial DLI cell dose and escalated-dose DLI regimen.6, 7, 10, 11, 12, 13, 14, 15 However, the majority of analyses regarding DLI in CML have been performed based on data of patients transplanted more than a decade ago, when most patients received a BM graft. Thus the results of treatment might have changed in the context of alloPBSCT in CML. It is conceivable, that the relapse of CML following PBSCT might be less susceptible to immune therapy with DLI than observed in previous studies, which were based on patients transplanted mostly with BM.
In order to provide an update on OS following DLI in the context of the stem cell source used for transplantation, we performed a retrospective analysis of data from the European Group for Blood and Marrow Transplantation (EBMT) registry regarding a total of 384 patients who relapsed after standard conditioning, HLA-identical sibling donor alloSCT performed in the years 1996–2005 who had been treated with DLI.
Patients and methods
This study was based on the registry of the EBMT and conducted within the Chronic Leukemia Working Party. Centers reported minimum essential data (MED-A) to a central database for all transplants performed, and many centers also reported a more comprehensive dataset (MED-B). The analysis has been performed in a group of 384 patients treated with DLI following PBSCT (N=168) and BMT (N=216). All the patients suffered from CML confirmed by the presence of the Ph chromosome, underwent standard intensity conditioning alloSCT, relapsed subsequently and were treated with DLI. The group was limited to patients transplanted from HLA-identical sibling donors in order to estimate the effects of interest more accurately in a homogeneous population.
Lymphocytes were collected from the donors by apheresis on one or more occasions and administered as single or multiple infusions. The phase of CML was classified in accordance with the criteria proposed by the Center for International Blood and Marrow Transplant Research.16 The advanced phases of CML were defined as all phases more advanced than CP1. Relapse was classified as molecular (that is, BCR-ABL transcripts detected by quantitative reverse transcriptase–PCR in two consecutive tests performed over a minimum of 4 weeks), cytogenetic (that is, reappearance of one or more Ph chromosome-positive metaphases at BM cytogenetics) or hematological (that is, presence of PB leukocytosis accompanied by a hypercellular BM with Ph chromosome on cytogenetic analysis) in accordance with previous reports.6, 17 Relapse stage at DLI was defined as relapse stage (molecular/cytogenetic/hematological) assessed on the date of first DLI infusion after first relapse or closest date before this infusion. The major end point of the study was OS after first DLI after first relapse after alloSCT, which was defined as the probability of survival since first infusion of DLI after first relapse irrespective of disease state at any point in time. Surviving patients were censored at last follow-up.
The baseline characteristics of the patients were summarized and tested with appropriate statistics for the different kinds of variables (t-test and Mann–Whitney U test for continuous variables and χ2 test and Cochran–Armitage test for trend for categorical variables). Survival curves were calculated according to the method of Kaplan and Meier18 and the different outcomes between the groups were univariately assessed by means of the log rank test. The multivariate analysis was performed by the Cox regression model. We focused on the impact of stem cell source allowing for differences between phase at SCT by including an interaction term and by including potential confounders (age at DLI, phase at SCT, donor–recipient sex match, interval SCT-relapse, interval relapse-DLI, DLI year, relapse stage at DLI). The confounders were selected on the basis of their significance in the univariate analyses and on their predictive impact as known from the literature. Hazard ratios (HRs) were estimated with 95% confidence intervals (95% CIs) and values of P<0.05 were considered statistically significant. We tested the proportionality assumption of all the predictors by means of a test based on scaled Schoenfeld residuals, which led to an improvement of the model by means of including time-dependent effects of the relapse status at DLI. All statistical analyses were performed using the statistical software SPSS (PASW Statistics versions 18.0 and 19.0, SPSS Inc. Chicago, IL, USA) and R version 2.12.2 with package ‘survival’ (http://www.r-project.org/foundation).
Characteristics of patients
The data were retrieved from the ProMISe database,19 and the selection included patients with CML who experienced disease relapse after standard intensity conditioning alloSCT from HLA-matched family donors and were treated with DLI. For the purpose of the current analysis, we selected only those patients who were transplanted either with PBSC or with BM. In order to avoid biases resulting from trends in stem cell source use in recent years, the group of patients was limited to those who were transplanted between 1996 and 2005, when a rather large number of transplants of each type was done each year. Based on the above criteria, data regarding 384 patients were retrieved. Patients’ detailed characteristics are shown in Table 1. In all 168 patients (44%) have been transplanted with PBSC, while 216 patients (56%) received BM. The analysis revealed that several features were not equally distributed between the groups. Patients after PBSCT were characterized by the more recent mean year of diagnosis (1999 vs 1998, P=0.007) and median year of alloSCT (2000 vs 1999, P<0.001) compared with patients after BMT. PBSC were also more frequently transplanted in advanced phases of the disease (23%) than BM (12%, P=0.009). In cases of PBSCT, the CMV mismatch was more frequently present (20%) than in cases of BMT (16%, P=0.021). T-cell depletion was also more frequently associated with PBSCT (59%) than with BMT (31%, P<0.001). The relapses occurred earlier after PBSCT (after median of 10.4 months) than after BMT (after median of 17.7 months, P<0.001). Although the median year of first DLI injection was the same for both groups (2001), the infusions were not equally distributed in years (P=0.055). Once the relapse was diagnosed, the first DLI was given after a significantly shorter median interval from relapse in the PBSCT arm (1.3 months) than in the BMT arm (2.9 months, P=0.014).
As the treatment with TKIs may affect the results of relapse treatment post-HSCT,20 we attempted to analyze the frequency of this event. Unfortunately, only limited data were available in the ProMISe database. According to this, 58 of the 262 patients (22%) for whom the information on TKI use was available have been treated with TKIs, and in 51 of them (19%), TKIs have been used after HSCT. The patients after BMT seemed to receive TKIs after transplantation more frequently (22.1%) than patients after PBSCT (15.2%) (P=0.17). Among 51 patients who received TKIs after HSCT in the general group, 34 started treatment at the time of first DLI infusion or later.
The major end point of this retrospective study was the OS after the first DLI course after relapse. The median follow up in survivors was 65.5 months from the first DLI infusion (range, 0.1–152.9 months). Of the 104 patients in whom death was observed, the median survival from the first DLI was 7.2 months (range, 0.2–110.5 months). The actuarial probability of survival was 82% at 1 year, 74% at 5 years and 70% at 7 years following the first DLI course (Figure 1, Table 2). The causes of death were classified as related to relapse or progression of the disease in 60 (63%) patients, as related to alloSCT in 21 (22%) patients and other in 14 (15%) patients. The data concerning the occurrence of GvHD following DLI was limited due to the retrospective nature of the study and low reporting rate regarding this event. The occurrence of ‘any type’ GvHD was reported for 44 patients of 108 with known data (41%). When the question was asked about the grade II–IV acute GvHD, it was reported for 10 of 156 (6%) patients. Data regarding chronic GvHD after DLI were not available.
When analyzed in the subsets of patients based on stem cell source used for transplantation, the estimates of OS in PBSCT arm were lower than in BMT arm (P=0.013, Table 2, Figure 1). There were no significant differences in the causes of death between PBSCT and BMT groups, with 65% vs 61% of deaths caused by relapse or progression of CML and 20% vs 25% of deaths caused by HSCT-related causes, respectively (NS). Relevant factors for OS in univariate analysis were: CML phase at the time of alloSCT (CP1 vs >CP1, P<0.001), interval between alloSCT and relapse (<12 months vs 12–24 months vs ⩾24 months, P<0.001), year when the first DLI was performed (⩽2000 or >2000, P=0.007) and stage of relapse at the time of first DLI infusion (molecular vs cytogenetic vs hematological, P<0.001) (Figure 1, Table 2). By contrast, the estimates of OS did not differ significantly between patients <40 and ⩾40 years of age at first DLI, with favorable or unfavorable configuration of sexes of donor and recipient and in those who received DLI within 2 months from the diagnosis of relapse or later.
In accordance with the current recommendations regarding differences in preferred stem cell source between patients transplanted in CP1 and in advanced phases,3 we performed a subset analysis of the influence of stem cell source on OS after DLI in these groups of patients. We did not observe any significant differences in OS after DLI between patients in CP1 transplanted with PBSC or BM (Figure 1, Table 2). By contrast, when analyzed in patients transplanted in advanced phases of CML, the estimates of OS were higher in the BMT group than in the PBSCT group, although the difference was NS.
In order to show the impact of stem cell source on OS following DLI corrected for other potentially confounding factors, we performed a multivariate analysis (Table 3). In this model, the stem cell source from PB appeared to have no significant impact on OS (HR 1.04, 95% CI 0.62–1.73, P=0.893) when the transplantation was performed in CP1. However, we observed a trend towards increased risk of death related to PBSCT in patients transplanted in >CP1 compared with less advanced stages. The HR for the interaction of advanced phase and PB was 2.18 (95% CI 0.93–5.09, P=0.073), translating into a HR of 1.04 × 2.26 × 2.18=5.09 for patients transplanted in advanced phase with PB vs patients transplanted in CP1 with BM (keeping the other predictors constant). Among the other factors affecting the OS following DLI, our analysis confirmed the negative impact of CML transplanted in advanced phase (HR 2.26, 95% CI 1.17–4.35, P=0.015) as well as more favorable OS in patients who relapsed after longer interval from alloSCT (HR 0.98 per month difference, 95% CI 0.96–0.99, P=0.007). We observed the trend towards better OS in patients who received first DLI after longer time from the diagnosis of relapse (HR 0.96 per month difference, 95% CI 0.91–1.01, P=0.09). Relapse type at first DLI was modeled with a time-dependent effect. The cytogenetic relapse type significantly decreased the probability of survival compared with molecular relapse type. HR at the time of DLI was 3.05 (95% CI 1.13–8.22, P=0.028), but the impact decreased over time (HR per year difference post-baseline 0.82, for example, leading to a HR at 2 years post-DLI of 3.04 × 0.822=2.07). The hematological relapse type had an even more pronounced effect on OS. The HR at time of DLI was 14.97 (95% CI 6.317–35.474, P<0.001); however, the risk related with this factor was also decreasing with time: HR per year difference post-baseline 0.637 (95% CI 0.470–0.863, P=0.004), for example, leading to a HR at 2 years post-DLI of 14.969 × 0.6372=6.07.
Once CML relapses after alloSCT, it can still be cured by cellular immunotherapy with DLI. The efficacy of this treatment was shown to be related to a number of factors listed in the Introduction section. Besides features of DLI itself (regimen, initial and total CD3+ cell dose), it seems to be affected by characteristics of the relapsed disease: the more aggressive, advanced-phase CML, higher tumor mass, the lower the efficiency of DLI. We hypothesized that the disease which survived and relapsed after a stronger GvL effect might be also less susceptible to DLI treatment because of a selection towards less immunosensitive cell clones. Therefore, the relapse of CML following PBSCT might be potentially less susceptible to cellular immunotherapy with DLI than in case of BMT, and this could result in reduced survival of patients.
Our univariate analysis revealed that the estimates of OS after the first DLI course were significantly lower in patients previously transplanted with PBSC than in those transplanted with BM. However, as shown in multivariate analysis, this effect was not solely related to stem cell source but rather with a number of factors associated either with PBSCT or BMT. Despite our efforts to achieve the highest homogeneity of the investigated group (selection of standard conditioning, matched family donors, treatment of relapse, selection of years with relatively comparable numbers of PBSCT and BMT), the groups differed in several respects. The fact that PBSCT was more frequently performed in advanced-phase CML could negatively affect the survival. Moreover, in the analyzed group of patients, the relapses after PBSCT occurred earlier than after BMT, while the early relapse is a negative prognostic factor and may be a consequence of more aggressive disease. The results could potentially be influenced by the treatment with TKIs after the relapse of CML. However, as the ProMISe database is focused on transplant-related events, it appeared that available data regarding treatment with TKIs were limited. Probably, therefore, an unexpectedly low rate of TKIs use was reported. Although the higher rate of post-HSCT TKIs use was expected in the PBSCT group (as performed more recently), our data revealed higher frequency of TKIs use in the BMT group. However, because of possible bias caused by low repoting rate, we cannot say if this could affect the above described differences in survival. Our analysis confirmed previous observations that advanced phase of CML at alloSCT, early relapse and advanced relapse (hematological>cytogenetic>molecular) are related with decreased OS after DLI.6, 11, 13 The importance of these as risk factors was also confirmed in multivariate analysis. A new finding was that the negative impact of cytogenetic or hematological relapse decreases with time. We have also shown for the first time that patients who were treated with DLI after the year 2000 had a significantly better survival than patients treated earlier. Although this observation has not been proven in multivariate analysis, it suggests that the efficacy of treatment with DLI improved in recent years.
The differential impact of stem cell source on outcomes of alloSCT in different CML phases stimulated us to analyze the outcomes of treatment with DLI in recipients of PBSC or BM who were transplanted for either CP1 or advanced disease. Although in patients transplanted in CP1 the differences in OS after DLI were negligible, patients after PBSCT in advanced-phase CML had decreased OS estimates compared with patients after BMT. However, the multivariate analysis confirmed the significance of the observation that the DLI used for treatment of CML transplanted in advanced phase with PBSC resulted in a decreased OS rate compared with other situations. The differences in response to DLI after PBSCT or BMT in patients transplanted in CP1 and >CP1 may be explained by the significant biological divergence between these CML phases. In the case of CP1 CML clearly responsive to immunotherapeutic approaches (as a characteristic feature of mature cells), the disease usually relapses in chronic phase and therefore the immunogenicity of the relapse may be conserved. In turn, the advanced phase CML, which usually relapses in advanced phase, contains more clonally evolved cells, which may proliferate more rapidly, are less immunogenic and therefore less responsive to immunotherapy. However, when advanced phase CML relapses after PBSCT, the GvL potential may have already been exhausted compared with BMT, where some effect may still be achieved with DLI. If this explanation were true, then the immediate recommendation would be to explore other treatment modalities other than DLI for patients relapsing after PBSCT if their CML had progressed to an advanced phase before the transplant. These alternative treatments may include second generation TKI, conventional chemotherapy or performing a second transplant given the relatively low efficacy of DLI shown in our analysis. Our observations may also support the recent trends towards reconsidering the role of BM as a source of HSCs for transplantation. The recent results of the randomized clinical trial of BMT vs PBSCT in hematological malignancies revealed comparable OS and leukemia-free survival after both approaches, with higher incidence of cGvHD and longer duration of immunosuppression in the PBSCT arm.21 Therefore, it might be wise to transplant CML patients with BM-derived cells limiting the incidence of transplant-related complications, keeping in mind that possible relapses may be more easily treated with DLI than in case of PBSCT.
The above analysis suffers from several weak points associated with retrospective collection of the data. It was performed using the EBMT registry, in which a number of important data were not available. Therefore, we do not have, for instance, data regarding occurrence of GvHD and BM aplasia after DLI. It should be noted, however, that the above analysis refers to a clinical situation, which is relatively rare, and that a prospective analysis might not be possible.
In conclusion, the stem cell source does not affect the OS of CML patients who underwent PBSCT in CP1, relapsed and were treated with DLI. However, when the patients were transplanted in advanced phases, previous PBSCT seems to negatively affect OS after DLI compared with BMT.
Gratwohl A, Heim D . Current role of stem cell transplantation in chronic myeloid leukaemia. Best Pract Res Clin Haematol 2009; 22: 431–443.
Baldomero H, Gratwohl M, Gratwohl A, Tichelli A, Niederwieser D, Madrigal A et al. The EBMT activity survey 2009: trends over the past 5 years. Bone Marrow Transplant 2011; 46: 485–501.
Stem Cell Trialists’ Collaborative Group. Allogeneic peripheral blood stem-cell compared with bone marrow transplantation in the management of hematologic malignancies: an individual patient data meta-analysis of nine randomized trials. J Clin Oncol 2005; 23: 5074–5087.
Schmitz N, Eapen M, Horowitz MM, Zhang MJ, Klein JP, Rizzo JD et al. Long-term outcome of patients given transplants of mobilized blood or bone marrow: a report from the International Bone Marrow Transplant Registry and the European Group for Blood and Marrow Transplantation. Blood 2006; 108: 4288–4290.
Eapen M, Logan BR, Confer DL, Haagenson M, Wagner JE, Weisdorf DJ et al. Peripheral blood grafts from unrelated donors are associated with increased acute and chronic graft-versus-host disease without improved survival. Biol Blood Marrow Transplant 2007; 13: 1461–1468.
Guglielmi C, Arcese W, Dazzi F, Brand R, Bunjes D, Verdonck LF et al. Donor lymphocyte infusion for relapsed chronic myelogenous leukemia: prognostic relevance of the initial cell dose. Blood 2002; 100: 397–405.
Collins RH, Shpilberg O, Drobyski WR, Porter DL, Giralt S, Champlin R et al. Donor leukocyte infusions in 140 patients with relapsed malignancy after allogeneic bone marrow transplantation. J Clin Oncol 1997; 15: 433–444.
Alyea EP, Soiffer RJ, Canning C, Neuberg D, Schlossman R, Pickett C et al. Toxicity and efficacy of defined doses of CD4(+) donor lymphocytes for treatment of relapse after allogeneic bone marrow transplant. Blood 1998; 91: 3671–3680.
Dazzi F, Hasserjian R, Gordon MY, Boecklin F, Cotter F, Corbo M et al. Normal and chronic phase CML hematopoietic cells repopulate NOD/SCID bone marrow with different kinetics and cell lineage representation. Hematol J 2000; 1: 307–315.
Elmaagacli AH, Basoglu S, Peceny R, Trenschel R, Ottinger H, Lollert A et al. Improved disease-free-survival after transplantation of peripheral blood stem cells as compared with bone marrow from HLA-identical unrelated donors in patients with first chronic phase chronic myeloid leukemia. Blood 2002; 99: 1130–1135.
Carlens S, Remberger M, Aschan J, Ringden O . The role of disease stage in the response to donor lymphocyte infusions as treatment for leukemic relapse. Biol Blood Marrow Transplant 2001; 7: 31–38.
Dazzi F, Szydlo RM, Cross NC, Craddock C, Kaeda J, Kanfer E et al. Durability of responses following donor lymphocyte infusions for patients who relapse after allogeneic stem cell transplantation for chronic myeloid leukemia. Blood 2000; 96: 2712–2716.
Raiola AM, Van Lint MT, Valbonesi M, Lamparelli T, Gualandi F, Occhini D et al. Factors predicting response and graft-versus-host disease after donor lymphocyte infusions: a study on 593 infusions. Bone Marrow Transplant 2003; 31: 687–693.
Kolb HJ, Schmid C, Buhmann R, Tischer J, Ledderose G . DLI: where are we know? Hematology 2005; 10 (Suppl 1): 115–116.
Simula MP, Marktel S, Fozza C, Kaeda J, Szydlo RM, Nadal E et al. Response to donor lymphocyte infusions for chronic myeloid leukemia is dose-dependent: the importance of escalating the cell dose to maximize therapeutic efficacy. Leukemia 2007; 21: 943–948.
Speck B, Bortin MM, Champlin R, Goldman JM, Herzig RH, McGlave PB et al. Allogeneic bone-marrow transplantation for chronic myelogenous leukaemia. Lancet 1984; 1: 665–668.
Chalandon Y, Passweg JR, Schmid C, Olavarria E, Dazzi F, Simula MP et al. Outcome of patients developing GVHD after DLI given to treat CML relapse: a study by the Chronic Leukemia Working Party of the EBMT. Bone Marrow Transplant 2010; 45: 558–564.
Matthews DE, Farewell VT . Using and Understanding Medical Statistics. Karger: Basel, 1996.
European Group for Blood and Marrow Transplantation Internet access: ProMISe. Available from www.ebmt.org/Contents/Data-Management/Registrystructure/ DatamanagementsystemProMIse/Pages/Data-management-system-ProMIse.aspx (Accessed on 21 September 2012).
Savani BN, Montero A, Kurlander R, Childs R, Hensel N, Barrett AJ . Imatinib synergizes with donor lymphocyte infusions to achieve rapid molecular remission of CML relapsing after allogeneic stem cell transplantation. Bone Marrow Transplant 2005; 36: 1009–1015.
Friedrichs B, Tichelli A, Bacigalupo A, Russell NH, Ruutu T, Shapira MY et al. Long-term outcome and late effects in patients transplanted with mobilised blood or bone marrow: a randomised trial. Lancet Oncol 2010; 11: 331–338.
The authors declare no conflict of interest.
About this article
Cite this article
Basak, G., de Wreede, L., van Biezen, A. et al. Donor lymphocyte infusions for the treatment of chronic myeloid leukemia relapse following peripheral blood or bone marrow stem cell transplantation. Bone Marrow Transplant 48, 837–842 (2013) doi:10.1038/bmt.2012.234
A Comparison of Long-Term Outcomes of Donor Lymphocyte Infusions and Tyrosine Kinase Inhibitors in Patients With Relapsed CML After Allogeneic Hematopoietic Cell Transplantation
Clinical Lymphoma Myeloma and Leukemia (2014)
Biology of Blood and Marrow Transplantation (2014)
Escalating-dose HLA-mismatched DLI is safe for the treatment of leukaemia relapse following alemtuzumab-based myeloablative allo-SCT
Bone Marrow Transplantation (2013)