Allogeneic stem cell transplantation (SCT) was considered to be the only curative therapy for patients with chronic myeloid leukemia (CML) until the introduction of the specific tyrosine kinase inhibitor, imatinib mesylate (IM), in 1999. In the 6-year update of the IRIS study, the estimated complete cytogenetic response (CCyR) and OS were 82% and 88%, respectively [1, 2]. However, results for patients in late chronic or advanced stages of the disease are inferior. For patients with an intermediate and high Sokal score, CCyR is 84% and 73%, respectively [3–6]. In addition, 25–30% of patients fail to achieve major molecular response (MMR) even after a long period of imatinib treatment and 10–15% of patients progress on treatment . The actual frequency of non-detectable BCR-ABL transcripts where molecular response might be considered "complete" is variable and ranges between 4% and 34% [5, 8–11]. Overall, about 18–30% of patients either progress or develop resistance or intolerance to imatinib [1, 12]. In the latest 7-year update of the IRIS study, imatinib was discontinued in 40% of patients due to intolerance (8%), lack of efficacy (15%), and various other reasons (17%) and was associated with a survival of 50% .
CCyRs in patients treated with second generation TKI after failing imatinib are much lower, ranging between 30 and 40% only [14–17]. The STIM study demonstrated a 61% relapse rate following IM cessation and apparently, a life-long need of treatment in most patients, reflecting the quiescent nature of the Philadelphia chromosome positive leukemic stem cell .
The decision to proceed to allogeneic SCT is largely based on consideration of the balance of risks. The European Bone Marrow Transplantation (EBMT) Risk Assessment Score has proved to be a valuable tool in predicting patient outcome [19, 20]. Attempts to reduce TRM and morbidity using TCD were largely unsuccessful due to an increased relapse rate [21–23]. We hereby present results of a cohort of 38 consecutive patients with CML, mostly in the preimatinib era, who were treated with allogeneic SCT using partial TCD and preemptive DLI, without post-transplant GvHD prophylaxis. We suggest that this method could be applied also in the imatinib era for patients who achieve a suboptimal response to TKIs, or for those who are referred to an allogeneic transplant based on a high risk of disease recurrence.
Patients and Methods
Thirty eight patients were enrolled in this nonrandomized study between March 1999 and May 2005 at the Rambam Health Care Campus (Haifa, Israel).The study was approved by the local Institutional Review Board and patients signed informed consent. Patients were considered eligible for the study if they had a matched related donor, Karnofsky score ≥80% and were in first chronic or accelerated phase. Imatinib was approved for the use in CML by the Israel Ministry of Health in April 2004. Only four patients with a low EBMT risk score entered the study after this date. Patient characteristics are shown in Table I.
Table I. Patient Characteristics
INF, interferon; Ara-C, cytosine arabinoside; SCT, stem cell transplant, EBMT, European Group for Blood and Marrow Transplantation; AML, acute myeloid leukemia; HLA, human leukocyte antigen.
Age : Median (range) years
Sex : male / female
Phase in transplant:
• 1st chronic phase
• 2nd chronic phase
• Accelerated phase
• Blast crisis
• INF ± Ara-C
Median time from diagnosis to SCT (range) months
• HLA identical sibling
• Mismatched family donor
EBMT risk score:
Conditioning regimen and transplant procedure
The conditioning regimen is represented in Table II. Oral busulfan was used and its final dose was individually determined based on measurements of serum busulfan area under the curve levels with a target dose of 850–1400 microM/minute. Transplants were performed in reverse isolation rooms equipped with HEPA filters. No post-transplant GvHD prophylaxis was given except for one patient who was transplanted from a partially mismatched family donor. Back-up marrow was harvested and cryopreserved for all patients. Post-transplant infection prophylaxis consisted of acyclovir, itraconazole, trimetoprim-sulfamethoxazole, and penicillin VK. Cytomegalovirus (CMV) status was determined weekly using PCR for CMV-DNA and pp65 antigenemia in blood leukocytes.
Table II. Stem Cell Transplantation Conditioning Regimen
ATG, anti-thymocytic globulin.
Day (−6 −4)
Day (−3 −2)
Day (−6 −2)
Day (−6 −2)
Donors were human leukocyte antigen (HLA) A,B,C serologically matched and DR and DQ molecularly matched siblings in 37 patients. One patient was transplanted from his father with a 9 out of 10 HLA molecular match. Donor stem cells were collected from peripheral blood using apheresis and following mobilization with 10 μg/kg/day G-CSF given subcutaneously for 5 consecutive days. CD34 cells were positively selected using anti-CD34 antibody conjugated to iron-dextran microbeads using CliniMACS device (Miltenyi Biotech, Bergisch Gladbach, Germany) with an aim to collect >5.0 × 106 CD34 cells/kg.
Definitions of study endpoints
Freshly isolated T cells at a final dose of 1 × 105 cells/kg were administered on the day of stem cell infusion. Doses of CD34 cells and T cells were determined using flow cytometry analysis and staining with monoclonal antibody (Becton Dickinson, Oxford, UK).
The day of neutrophil engraftment was defined as the first of 3 consecutive days with an absolute neutrophil count (ANC) >0.5 × 109/l. The day of platelet engraftment was defined as the first day of a platelet count >20 × 109/l without platelet transfusion on 5 previous days.
Patients who failed to engraft by Day 30 received either backup marrow or a second transplant from the same donor.
Acute GvHD (aGvHD) was graded according to standard criteria . Only patients surviving for more than 30 days were included in the aGvHD analysis. Chronic GvHD (cGvHD) was assessed in patients who were alive 90 days after transplant.
TRM was defined as the cumulative incidence of death occurring as a result of the transplant with no evidence of relapse.
LFS was measured from the date of transplantation to treatment failure (defined as hematological, cytogenetic, or molecular relapse), or death from any cause; otherwise patients were censored on the date of the last follow-up.
OS was measured from the date of transplantation until death from any cause, censoring patients alive on the date of the last follow-up.
Following transplant all patients were under close surveillance for the presence of minimal residual disease (MRD), using cytogenetic analysis and PCR for detection of BCR-ABL transcripts. Bone marrow and peripheral blood samples were analyzed every 3 months during the first year post transplant and every 3–6 months in subsequent years.
PCR method: RQ-PCR was performed according to the Europe Against Cancer (EAC) protocol , with a RQ-PCR sensitivity of 10−5, as part of the UK NEQAS program .
Definition of disease status
Hematological, cytogenetic and molecular responses post transplant were determined according to the European LeukemiaNet expert panel guidelines .
Molecular response was defined as complete (CMR) if no BCR-ABL/ABL transcripts were found by RQ-PCR and major molecular response (MMR) was defined as BCR-ABL/ABL transcript ratio ≤0.1% of baseline at three consecutive analyses performed at 4-week intervals.
Hematological relapse was diagnosed using standard hematological criteria. Cytogenetic relapse was established if Ph+ metaphases were detected after a period of complete cytogenetic remission (CCR). Molecular relapse was diagnosed if BCR-ABL/ABL transcripts were detected by RQ-PCR after a period of nondetection or a rise of ≥1 log in quantitative assay.
Donor leukocyte infusion
Donor leukocyte infusions (DLI) were delivered in the escalating dose starting at 3 × 106 CD3 cells/kg followed, if necessary, by 1 × 107 cells/kg, 5 × 107 cells/kg and 1 × 108 cells/kg.
DLI was administrated upon hematological or cytogenetic relapse and in case of persistence/reappearance of BCR-ABL transcripts as detected by RT-PCR or increase of ≥ 1 log by RQ-PCR, starting from 6 months post-transplant onward. In cases when more than 1 DLI were given, the successive escalated dose was given at ≥3-month intervals as dictated by MRD follow-up.
Descriptive statistics were performed in order to calculate means, confidence intervals, medians, and frequencies of different variables.
Cumulative survival, LFS rates and time to progression were computed using the Kaplan Meier product limit method where patients were censored at the latest follow-up if still alive.
Stem cell harvest and engraftment
Following cell separation with the CliniMACS device, the median CD34+ cell dose collected was 9 × 106 cells/kg (range, 5–24) with 4 logs of T cell depletion. Median CD34 + cell purity was 97%.
Of the 36 patients alive beyond 30 days post transplant, 35 (97.2%) engrafted. The median time for neutrophil engraftment was 11 days (range, 9–25) and 15 days for platelet engraftment (range, 9–134). One patient did not engraft and received backup autologous marrow. Two patients (5%) experienced late graft failure at Days 35 and 40 post transplant. One was salvaged by stem cell boost from the original donor and the other received stored back up marrow. Engraftment was successful in all these three patients.
With a median follow up of 90.5 months (range, 1-134), 32 of 38 (84%) patients are alive. Two patients died from transplant-related complications. The non-relapse mortality (NRM) beyond day 100 was 8%. These patients died from the development of late autoimmune complications at 6–14 months post transplant. The cumulative incidence of death from transplant was 13% (Fig. 1).
Acute and chronic GVHD
Despite the absence of post-transplant GvHD prophylaxis, there was no mortality associated with GvHD. Seven patients (18%) developed aGvHD which was Grades I–II in 5 of 7. None of the patients developed Grade IV aGvHD and all responded well to a short course of immunosuppressive treatment. cGvHD developed in only one patient on Day 264 (acute type) and resolved completely after cyclosporine treatment.
The 4 log reduction in T cells did not translate into unusual infectious complications. Disseminated Herpes Zoster occurred in two patients (one of whom was on immunosuppressive therapy for aGvHD), 10 and 20 months post transplant. Pneumonia caused by Legionella and Mycobacterium tuberculosis developed in two patients. All patients recovered fully following the appropriate treatment. None of the patients developed CMV disease.
Five patients (13%) developed late immune complications which included idiopathic thrombocytopenic purpura (ITP) (one patient), autoimmune hemolytic anemia (AIHA) (one patient), ITP combined with AIHA (one patient) and systemic vasculitis (one patient). These complications appeared between 1.5 and 6 months post-transplant and were fatal in three patients.
Donor lymphocyte infusion
Twenty-four patients (63%) received DLI. Median time to first DLI was 9 (range, 6–60) months. Nine patients received two doses with a median time to second dose of 21 (range, 6–98) months and four subjects received ≥3 doses. DLI was given to 12 patients for persistence or relapse of cytogenetic disease and to another 12 patients for persistence of molecular disease. Following DLI, seven patients developed aGvHD which was Grades I–II in 5. None developed Grade IV aGvHD and there was no mortality associated with DLI. Four patients developed cGvHD which was limited in two patients and extensive in the other 2. Only one patient is currently on immunosuppressive treatment. All patients responded to DLI with either MMR or CMR. Figure 2 represents the kinetics of PCR for BCR-ABL response to DLI.
Three patients developed blast crisis (1 myeloid and 2 lymphoid) 8-34 months post transplant. Two patients (myeloid in 1, lymphoid in 1) responded to combined leukemia induction chemotherapy followed by DLI and imatinib and one patient died from sepsis while receiving induction.
Five patients are currently on imatinib, two for a previous blast crisis, one due to donor nonavailability for DLI collection and two patients who received backup marrow for nonengraftment.
Four patients entered transplantation beyond the first chronic phase. All of them are currently in CMR which was achieved in three patients following DLI.
Thirty-two patients are alive with a median follow up of 90.5 months. Of these, 100% have ≥ MMR and 84.4% of patients are in CMR. Overall 5-year LFS is 78.95%. Median LFS has not been reached. Kaplan-Meier curve of LFS is shown in Fig. 3A. 91.5% of patients are leukemia free, with a 5-year OS of 84.2. Median survival time has not been reached. Kaplan Meier curve of OS is demonstrated in Fig. 3B.
The introduction of the TKI, imatinib, has revolutionized the treatment of CML and currently it has become first line therapy for patients in chronic phase. Accordingly, the number of transplants performed for CML decreased dramatically to ∼4% of all transplants done in Europe [20, 28].
Allogeneic SCT, although leading to a cure in a significant number of patients, is a complex procedure associated with high morbidity and mortality. The EBMT risk score has identified the risk factors and the probability of OS, LFS, and TRM in a large cohort of patients following transplant . These data were confirmed in several independent series .
Good-risk patients with a risk score of 0-1 were reported to have OS, LFS and NRM of 74.5%, 61.5%, and 21%, respectively . In the current study, the 5-year OS and LFS as well as NRM in patients with a score of 0-1 were 93%, 85.7%, and 0%, respectively. Notably, a superior outcome was also evident in patients with a higher EBMT risk score (Table III), although we recognize that the numbers were small when subdivided by risk categories.
Table III. Comparison of Transplant Outcome Across Risk Factors to EBMT Results
NRM, nonrelapse mortality; LFS, leukemia-free survival; OS, overall survival. The numbers in the table represent results of the current cohort; the numbers in brackets represent the EBMT results  for the corresponding risk score.
Both LFS and OS were calculated at 5 years.
Because of the progress in supportive care, significant improvement was reported in the outcome of CML patients transplanted over the last decade. However, such outcome cannot be attributed to this factor alone, since the 1-year TRM and OS in our study were 13% and 84%, compared with 30% and 61%, respectively reported in a more recently transplanted cohort . Significance of the results is further emphasized by low probability of acute and chronic GvHD, even following DLI, without compromising LFS and with no increase in relapse rate.
In an attempt to decrease TRM and GvHD, while increasing LFS and OS, various groups described their experience with T cell depletion (TCD), reduced intensity transplantation and DLI.
The use of negatively selected stem cells followed by bulk administration of DLI to patients with hematological or cytogenetic relapse [21, 22] resulted in a decrease in acute and chronic GvHD. Remarkably, relapse rate appeared to be significantly higher in the TCD group compared with the non-TCD cohort. Ex vivo TCD using Campath −1G followed by graded DLI was reported in a small cohort of patients, 78% of whom relapsed, with 91% entering CR following DLI . Elmaagacli et al. presented a retrospective analysis comparing a non-TCD transplant with the transplant utilizing positive selection of CD34 cells uniformly followed by T cell addback on day 120 post-transplant, regardless of disease status, with favorable estimated 3-year OS (68% versus 90%, respectively) .
In our study, partial instead of complete T-cell depletion was used for the following reasons: it enabled engraftment in patients with myeloproliferative disease which is known to be associated with a relatively high incidence of graft rejection, contributed to preservation of immunity to infectious diseases and promoted disease control in at least part of the patients. The reduction in T-cell dose allowed to avoid immunosuppressive therapy with its inherent toxicity and reduce the risk of significant GvHD.
DLI is an effective method to restore remission in patients relapsing after transplantation. It can be given as a “bulk” in a single dose or in escalated doses and is usually administered with or without imatinib if there is evidence of disease relapse [30–33]. Preemptive DLI, dictated by patients' PCR and/or cytogenetic status, has the advantage of treating only patients prone to relapse at MRD and avoiding unnecessary therapy in a substantial number of patients. In addition, the use of graded DLI contributed to decrease in GvHD, the major cause of morbidity in allogeneic SCT. Furthermore, DLI administration at MRD maximizes the potential of this immunotherapeutic strategy to be curative . In a recent survey of 328 patients receiving DLI after BMT for CML, 38% developed GvHD following DLI; this was associated with a 2.3-fold increase in the risk of death. These findings are in contrast to our data where no mortality was associated with DLI.
The major shortcoming of this study was the occurrence of various autoimmune complications leading to increased late mortality. Post-transplant immune complications (specifically AIHA) are reported in 3–19% of transplants [35, 36]. In various studies, immune complications were associated with TCD, umbilical cord transplantation and transplantation for non-malignant diseases. The reported mortality in these series approaches 80%. This complication is thought to originate from unbalanced reconstitution of B & T cell lymphopoiesis post-transplant and absence of regulatory clones which may enable proliferation of pathological clones. The outcome in our cohort matches that of the published reports [35, 36].
Although imatinib is the initial treatment for CML patients, individuals with poor response according to the European LeukemiaNet  recommendation or intolerant to tyrosine kinase inhibitors should be offered allogeneic transplantation. Indeed, a recent report demonstrated that LeukemiaNet criteria have a useful predictive value for disease outcome in suboptimal responders to TKIs and that these patients do far less well . The update of management recommendations by the European LeukemiaNet  delineates hematological and cytogenetic resistance at earlier time points compared with initial recommendation and provides new definitions of suboptimal response and failure following treatment with second generation TKIs. Some investigators even debate using second generation of TKIs in patients failing imatinib and consider transplant as best choice [39, 40]. Current perception in the management of CML is aimed at achieving CMR. Using IM, the CMR rate approximates 20%, with a 61% relapse rate in patients discontinuing IM, thus leaving allogeneic SCT as a valid option for appropriately selected patients.
In this regard, the German CML IV multicenter study demonstrated a 91% 3-year survival in transplanted patients with a low EBMT risk score and a high-risk disease. In a matched pair comparison there was no survival difference between transplanted and nontransplanted patients and the authors conclude that allogeneic transplantation could become the second choice in selected patients . In addition, recent data on SCT results in patients with advanced phase CML (≥CP) resistant to imatinib, demonstrated an OS of only 44% and event-free survival (EFS) of 36% at 2 years . It can also be argued that young patients with an HLA-matched sibling donor and a low EBMT risk score could still be offered allogeneic SCT instead of lifelong treatment with TKIs , especially using low-risk procedures, such as those described above. Another group of patients that may be offered an early transplant are those inhabiting regions where access to TKIs is limited due to economic reasons. Early transplantation would be a preferred choice for patients with a low EBMT risk score .
In a disease such as CML, given its initial chronic nature, the sole use of partial T cell depletion allowing for generation of tolerance on the one hand and GVL effect on the other hand, without impairing immune reconstitution, may be a perfect platform for exploiting the maximum benefit of GVL at a minimum cost in terms of toxicity. Application of this risk-adapted approach has been translated into significantly improved long-term OS and LFS.
The authors would like to thank Miltenyi Biotec for the generous support of this research and Sonia Kamenetsky for assistance in preparation of the manuscript.
Author Contributions: TZ was the principal investigator, wrote the paper and takes primary responsibility for the paper. TK interpreted the data. NH recruited the patients. RF recruited the patients. EJD recruited the patients. IA recruited the patients. YO recruited the patients. IG recruited the patients. TF recruited the patients. DS interpreted the data. EH interpreted the data. ES interpreted the data. YR interpreted the data. JMR wrote the paper.