Dose-reduced conditioning for allografting in 44 patients with chronic myeloid leukaemia: a retrospective analysis


  • Martin Bornhäuser,

  • Michael Kiehl,

  • Wolfgang Siegert,

  • Johannes Schetelig,

  • Bernd Hertenstein,

  • Hans Martin,

  • Rainer Schwerdtfeger,

  • Herbert G. Sayer,

  • Volker Runde,

  • Nikolaus Kröger,

  • Catrin Theuser,

  • Gerhard Ehninger,

  • the Cooperative German Transplant Study Group

Dr Martin Bornhäuser, Medizinische Klinik und Poliklinik 1, University Hospital Carl Gustav Carus, Fetscherstrasse 74, 01307 Dresden, Germany. E-mail:


This retrospective study describes the outcome of patients with chronic myeloid leukaemia after allografting using dose-reduced conditioning with fludarabine and busulphan. Forty-four Philadelphia chromosome (Ph)-positive patients were transplanted in nine German centres; 26 patients were in chronic phase, 11 in accelerated phase and seven in blast crisis. Thirty-four patients achieved complete remission, with 18 alive and disease-free at a median follow-up of 562 d (range 244–922 d). Grade II–IV acute graft-versus-host disease (GVHD) incidence was 43%. Twenty patients died, 15 of causes unrelated to relapse. Risk factors predisposing to graft failure by univariate analysis were an unrelated donor (8/23 compared with a related donor 2/21, P = 0·07) and interferon therapy within 90 d of transplant (4/6 versus 3/17, P = 0·025). At the last follow-up, of 31 patients for whom molecular or cytogenetic data were available, 16 (52%) were polymerase chain reaction-negative, and seven (23%) were Ph-negative by fluorescent in situ hybridization. These findings demonstrate that dose-reduced conditioning with fludarabine and busulphan provides durable engraftment and a low rate of relapse. However, in this population, many of whom were not eligible for high-dose conditioning due to age, reduced performance status, previous complications or extensive pre-treatment, these data highlight the need for effective anti-infectious and GVHD prophylaxis. In addition, this study supports the discontinuation of interferon therapy at least 90 d before transplant

High-dose conditioning before allografting has been used to control malignancy and prevent graft rejection, but is also associated with a high incidence of acute and long-term side-effects. Transplant-related mortality may reach 50% in patients over 45 years of age (Gratwohl et al, 1993) and this has prevented the use of these transplant protocols in patients over the age of 55 years with poor performance status, or those who have been heavily pre-treated.

Dose-reduced conditioning aims to reduce the malignant clone without complete myeloablation (Slavin et al, 1998; Carella et al, 2000a). The graft-versus-leukaemia (GVL) effect of the allograft can then be harnessed to complete the eradication of malignant cells, which, in chronic myeloid leukaemia (CML), are particularly susceptible to attack by donor lymphocytes (Kolb et al, 1995; Porter & Antin, 1999). Reduced-dose conditioning using drugs such as fludarabine and busulphan allows transplant protocols to be extended to older patients and to those who are not considered eligible for high-dose therapy (Slavin et al, 1998). It is hoped that such procedures will reduce morbidity and mortality caused by post-transplant infection, and also reduce the incidence of graft-versus-host disease (GVHD). Encouraging results have been obtained in treating lymphoma and leukaemias (Khouri et al, 1998; Grigg et al, 1999; Carella et al, 2000b; Nagler et al, 2000) including success in a small number of CML patients (Childs et al, 1999; Gomez-Almaguer et al, 2000).

Allografting and interferon-α (IFN-α) therapy are the two treatments of choice for CML. Long-term application of IFN-α can produce and sustain remissions but, because total eradication of the malignant clone is not achieved, the ultimate outcome is relapse and disease progression (Hasford et al, 1998). Allografting can achieve true molecular remission but is associated with some risk, even when a matched sibling donor is available. Therefore identifying factors that influence and may benefit patient outcome is of prime importance in optimizing this treatment option.

This retrospective, multicentre study analysed the results of 44 allogeneic transplants after dose-reduced conditioning with the aim of identifying factors that could predict for graft failure.

Patients and methods

Patients Forty-four patients with CML who underwent allogeneic transplantation at nine German centres after dose-reduced conditioning with fludarabine and busulphan were included in this analysis. Twenty-seven patients were men and 17 were women. The median age for the whole group was 52 years (range 25–65 years). At the time of transplant, 26 patients were in chronic phase (CP), 11 in accelerated phase (AP) and seven in blast crisis (BC). All patients were Philadelphia chromosome-positive (Ph), and most were considered high risk for conventional conditioning because of age, reduced performance status, failure of previous allograft, previous complications or extensive pre-treatment. The median interval from diagnosis to transplantation was 26 months (range, 3–125 months).

The median number of therapies before conditioning was two (range 1–5). The majority of patients (84%) had previously received hydroxyurea (n = 37, Table I). Of the 24 patients who had received IFN-α and for whom response data was available, two had major responses, four had minor responses and eight failed to respond.

Table I.  Previous treatment regimens.
  1. DAV, daunorubicin, cytarabine, etoposide; ICE, idarubicin, cytarabine, etoposide; HAM, High dose cytarabine, mitoxantrone; TAD, thioguanine, cytarabine, daunorubicin; FLAG, fludarabine, cytarabine, G-CSF; CLAEG, cladribine, cytarabine, etoposide, G-CSF.

Carboplatin, VP16, FLAG, myelobromol, 6-MP, vincristine, vinblastin, cyclophosphamide, mitoxantrone, CLAEG1
Previous allogeneic transplant3
Interferon therapy, n24
Duration (months) median (range)12 (1–60)
Interval before transplant (months) median (range)7 (1–78)

Conditioning regimens Pre-transplant, patients received fludarabine for 5 or 6 d and busulphan was given for 2 d (Table II). Thirty-four patients received antithymocyte globulin (ATG) as an immunosuppressive agent over the 4-d period before transplant.

Table II.  Conditioning regimens.
Fludarabine (dose)
(dose mg/m2)
(dose mg/kg)
(dose × 100 mg/m2)
125–1808 (po) 25 (56·8)
150–1806·6 (iv) 12 (25)
2408 (po) 3 (6·8)
125–18012–16 po/iv 3 (6·8)
1508 (po)91 (2·3)

Graft characteristics HLA-matched grafts from related donors were used in 19 transplants (43%), and from unrelated donors in 15 transplants (34%). Grafts from related donors mismatched at two and three HLA sites were used in two transplants (4·5%), grafts from unrelated donors mismatched at one HLA site in six transplants (13·6%), whereas grafts from unrelated donors mismatched at two sites were used in two cases (4·5%). Grafts were matched for sex in 23 cases (52·3%), female recipients received male grafts in 12 cases (27·2%) and male recipients received female grafts in nine cases (20·5%). Stem cells from unmanipulated bone marrow were used in 15 cases (34·1%) and from granulocyte colony-stimulating factor (G-CSF) mobilized peripheral blood in 29 cases (65·9%).

GVHD Acute GVHD was graded clinically (Przepiorka et al, 1995), and confirmed by biopsy when necessary. GVHD prophylaxis consisted of either oral or intravenous cyclosporine (CSA), adjusted to achieve trough blood levels of 200–300 ng/ml in 21 cases, with the addition of 2 × 1 g mycophenolate mofetil (MMF) in 13 cases (29·5%), 10 mg/m2 methotrexate in nine cases (20·5%) and 100 mg steroids in one case (2·3%). One patient (2·3%) received MMF alone.

Molecular analysis Chimaerism analysis was performed at various time points using polymerase chain reaction (PCR) amplification of short tandem repeat (STR) markers as described by Thiede et al (1999). Fluorescent in situ hybridization (FISH) and PCR to detect BCR–ABL were performed using standard techniques (Kiss et al, 1999; Najfeld et al, 1997).

Statistical analysis Overall survival, disease-free survival and probability of relapse were calculated according Kaplan & Meier (1958) at a median follow-up duration of 562 d (range 244–922 d). Only patients with documented engraftment for more than 1 month without signs of persistent disease were evaluated for disease-free survival. Univariate analysis of risk factors for graft failure was carried out using Fischer's exact test. Cox proportional hazards analysis was used to compare factors influencing engraftment and GVHD. Risk assessment according to Gratwohl et al (1998) assigned the following Gratwohl scores; two patients scored 1; two scored 2; 10 scored 3, 18 scored 4, nine scored 5 and three scored 6.


The median number of CD34+ stem cells infused was 4·32 × 106 cells/kg (n = 36, range 0·96–19·1 × 106 cells/kg). Transfusion requirements were relatively low, with a median of 6 units of packed red blood cells (RBC) transfused (range 0–18 units), and a median of 3 thrombopheresis units transfused (range 0–24 units) per patient. Neutrophil counts remained lower than 0·5 × 109/l for a median of 7·5 d (range 0–45 d) and platelet counts remained lower than 20 × 109/l for a median of 5 d (range 0–45 d) post-transplant. Primary neutrophil engraftment [>0·5 × 109/l polymorphonuclear (PMNC)] occurred at a median of 14 d post-transplant (range 7–38 d), whereas platelets recovered to > 20 × 109/l at a median of 17 d post-transplant (range 9–113 d). Engraftment failed in only four cases (9·1%) and late graft failure occurred in six cases (13·6%). Donor chimaerism was complete in almost all cases and achieved at a median of d 52 (range 12–545 d), with a median maximum recorded value of 98·8% (range 2·8–100%).

Univariate analysis demonstrated that graft matching as a result of donation from a relative promoted engraftment more often than matching by sex chromosomes or at human leucocyte antigen (HLA) sites, although significance values were borderline (Table III). IFN-α therapy within 90 d before transplant also adversely affected engraftment at low levels of significance. Use of ATG had no significant effect nor did CD34+ dosage, although CD34+ counts were not available for four of the patients with graft failure. Multivariate analysis failed to reveal significant factors for graft failure, partly because of the small size of the groups analysed.

Table III.  Transplant characteristics and relationship to graft failure by univariate analysis.
CharacteristicGraft failureP
  • *In four cases of graft failure CD34+ counts not available

  • **

    In two cases status not known.

  • ATG, antithymocyte globulin; BM, bone marrow; IFN-α, interferon-α; PBSC, peripheral blood stem cells; NS = not significant.

Related donor versus unrelated donor2/21 versus 8/230·07
ATG versus no ATG16/33 versus 3/11NS
PBSC versus BM5/29 versus 5/15NS
CD34+ counts > 4·32 × 106/l versus < 4·32 × 106/l4/18 versus 2/18*NS
IFN-α < 90 d versus > 90 d4/6 versus 3/170·025
IFN-α versus no IFN-α7/24 versus 2/9**NS
Sex mismatch versus match5/20 versus 4/22NS
HLA mismatch versus match1/8 versus 9/36NS

A total of 74% patients (21 of 31) with data available had durable engraftment as demonstrated by molecular analysis of peripheral blood or cytogenetic analysis of bone marrow. At the last follow-up, 16 patients (52%) were PCR-negative and seven patients (23%) were Ph using FISH. In addition, three patients (6·8%) became PCR-negative immediately after transplant but were positive at the last follow up.

GVHD and infectious complications

Acute GVHD was observed in 27 patients; grade I (n = 8), grade II (n = 13), grade III (n = 5) and grade IV (n = 1). The rate of acute GVHD (grades II–IV) did not vary whether CSA was used alone or in combination (10/20 versus 10/20).

Complications arose in 22 cases (50%) within the first 30 d after transplantation. Most of these were caused by infections, including pneumonia (n = 7), bacterial infection (n = 7) and fungal infection (n = 3).

Relapse and survival

Eighteen patients (41%) are alive and disease free at a median of 562 d (range 244–922 d) after transplantation. Median overall survival for the whole cohort was 450 d (range 9–922 d).

Patients in CP were less likely to relapse compared with patients in BC (P = 0·016), as determined by subgroup analysis of patients with respect to disease stage at transplant (Fig 1), although, due to transplant-related mortality, this did not translate into longer overall and disease-free survival (Fig 2). Twenty patients died, 15 of causes other than relapse. The causes of death were as follows: fungal infection, three; lung toxicity, three; hepatic toxicity, one; cardiac toxicity, one; multiple organ failure, two; encephalitis, one; sepsis, one; haemorrhage, one; GVHD, one; unknown causes, one.

Figure 1.

Probability of relapse according to disease stage at transplant.

Figure 2.

(A) Overall survival (OS) according to disease stage at transplant. (B) Disease-free survival (DFS) according to disease stage at transplant.


This retrospective analysis confirms the ability of dose-reduced conditioning using fludarabine and busulphan before allografting to produce rapid and durable engraftment with low relapse rates for patients with CML (Bornhauser et al, 2000). In addition, the data cautiously support the discontinuation of IFN-α therapy at least 90 d before conditioning.

Relapse-free survival data indicated a dependence on the stage of disease at transplant, with patients in BC or AP showing higher relapse rates than patients in CP. However, these results did not translate into equivalent differences in overall survival, and treatment-related mortality played a more significant role than relapse.

The incidence of acute GVHD was high in these patients, possibly due to older age or to the use of CSA only as prophylaxis in about half of the patients (Przepiorka et al, 1999). Despite a rapid recovery of neutrophils and platelets, there were a number of complications caused by infections, indicating that rigorous prophylaxis may be required for this patient group. In part, this may relate to the age and relatively poor performance status of patients, and the fact that most patients were heavily pre-treated. Taking into account donor type, disease stage, age of recipient, sex matching of graft, and time from diagnosis to transplantation (Gratwohl et al, 1998), most patients achieved Gratwohl scores of 3, 4 and 5. These scores predict a risk of transplant-related mortality of 31%, 46% and 51% respectively. Overall, transplant-related mortality was 36% in this group of patients and, although this is lower than has been recorded with conventional therapy, particularly high-dose chemotherapy and total body irradiation, morbidity and mortality remain high compared with conventional allografting. Future prospective studies in which reduced-dose conditioning is compared directly with conventional regimens are needed to establish the importance of conditioning in the risks accompanying allogeneic transplantation.

The use of an unrelated donor versus a related donor was an unfavourable risk factor for graft failure. HLA mismatching, however, did not significantly reduce engraftment success. This is consistent with work in murine models indicating that optimized immunoablative fludarabine-based preparative regimens can prevent rejection of fully major histocompatibility complex (MHC)-disparate marrow (Petrus et al, 2000). Factors such as immunodepletion with ATG or sex matching of graft and recipient did not significantly influence engraftment. The use of mobilized peripheral blood stem cells was as successful in promoting engraftment as the use of unmanipulated bone marrow. CD34+ dose above and below the median value of 4·32 × 106 cells/kg was not found to be a significant factor in graft success. Previously, the infusion of more than 3·12 × 106 CD34+ bone marrow cells/kg was suggested to improve the outcome after allogeneic transplantation of CML patients (Morariu-Zamfir et al, 2001), a dose above the minimum threshold (2 × 106 cells/kg) was suggested by other work (Singhal et al, 2000).

Administration of IFN-α within 90 d of transplant was also significantly associated with graft failure, although it is controversial whether previous IFN-α treatment can compromise transplantation. Previously, long-term use of IFN-α has been implicated in poorer outcome in allografting for CML (Beelen et al, 1995; Morton et al, 1998), associated with an increase in GVHD and a decrease in survival in the 6-month period after transplantation. However, other studies failed to show a link with GVHD or post-transplant survival (Giralt et al, 1993; Tomas et al, 1998; Zuffa et al, 1998). More recently, the timing of IFN-α treatment was related to patient outcome, with the suggestion that adverse effects can be minimized by discontinuation of IFN-α at least 90 d before transplantation (Hehlmann et al, 1999), which is in agreement with our data. In addition, short-term IFN-α application has now been implicated in graft failure in a large-scale retrospective study of 854 patients undergoing HLA-matched sibling allografting in CML (Giralt et al, 2000).

Although the patient numbers were small, this analysis after reduced-dose conditioning indicates that the use of IFN-α within 90 d of transplant may increase graft failure, whereas outside this period it can be used with relative safety. IFN-α should therefore be used with caution in patients who are eligible for allografting. As the continuing refinement of dose-reduced conditioning regimens widens the range of patients eligible for allografting, findings such as these becomes increasingly important.

While overall survival remains acceptable in this high-risk population, many of whom had failed attempts to use IFN-α to stave off AP, the predominance of transplant-related deaths over deaths caused by relapse, highlights the need for more effective anti-infectious and GVHD prophylaxis. New therapies are also emerging that offer hope for CML, and which may eventually supersede IFN-α as initial therapy and influence the success of allografting. These therapies include STI-571, which targets and inhibits the oncogenic Bcr-Abl kinase so eliminating the Ph+ clone, is currently showing promise in inducing cytogenetic remission in stage 1 trials, and may ultimately alter perceptions of the best time to initiate allogeneic transplantation.


The authors would like to thank all of the physicians and nursing staff in the participating study centres who have been involved in the care of the patients.