It's moving day: factors affecting peripheral blood stem mobilization and strategies for improvement


Dr Stefan Fruehauf, Department of Internal Medicine V, University of Heidelberg, Hospitalstrasse 3, D-69115, Heidelberg, Germany. E-mail:

Mobilized peripheral blood stem cells (PBSC) have become the main source for autologous or allogeneic transplantation following myeloablative therapy in patients with haematolymphoid malignancies or solid tumours. Classical strategies for PBSC mobilization include administration of granulocyte colony-stimulating factor (G-CSF) alone or in combination with other cytokines or myelosuppressive chemotherapy. PBSC mobilization and collection have been optimized in numerous clinical trials, but a significant proportion of patients fails mobilization. In recent years, some of the underlying physiology has been elucidated, leading to the development of new mobilization strategies. Expression of the G-CSF receptor on stem cells is not required for mobilization. G-CSF or myelosuppressive therapy act via secondary pathways, including the chemokine stromal-derived factor-1 and its receptor CXCR4. New chemokine receptor agonists lead to a rapid and substantial PBSC mobilization. The most advanced data are available for AMD-3100, which acts synergistically with G-CSF in normal donors. This CXCR4 ligand appears to mobilize more primitive PBSC with increased repopulation capacity when compared with G-CSF. Wider recognition of these new developments may enable sufficient PBSC to be obtained for transplantation in poorly mobilizing patients, thus enabling them to receive potentially curative treatment.

The success story of pbsc transplantation

In the mid 1980s, PBSC became a source for transplantation and the ‘modern era of stem cell transplantation’ began (Juttner et al, 1985; Kessinger et al, 1986; Körbling et al, 1986, Reiffers et al, 1986). In the early 1990s, PBSC replaced bone marrow as the preferred source because of the relative ease of collection, the faster haematological recovery when compared with bone marrow transplantation (BMT), leading to fewer complications, and lower costs of the procedure (To et al, 1997). Furthermore, several studies have shown that the overall survival and disease-free survival was superior in patients that received PBSC compared with those that underwent BMT (Heldal et al, 2000; Bensinger et al, 2001), although other groups did not find a difference in survival (Vigorito et al, 1998; Blaise et al, 2000). By 1998, 95% of all autologous transplants and 25% of all allogeneic transplants used PBSC (Goldman & Horowitz, 2002). The number of allogeneic PBSC transplants has further increased in recent years.

Despite the success of PBSC transplantation, the exact mechanisms involved in PBSC mobilization and homing are not yet clear. In adults, small amounts of PBSC are present in the peripheral blood (PB) during steady state haematopoiesis, as has been known for 40 years (Goodman & Hodgson, 1962; Dorie et al, 1979; Fruehauf et al, 1995), suggesting a continuous migration and exchange of haematopoietic stem cells between the bone marrow (BM) and other organs, such as the liver and spleen (Starzl et al, 1996; Nierhoff et al, 2000). This model is strongly supported by experiments in which parabiotic mice were surgically conjoined and shared a common circulation (Wright et al, 2001). They rapidly established stable, functional cross-engraftment of partner-derived haematopoietic stem cells (HSC) and maintained partner-derived haematopoiesis after surgical separation (Wright et al, 2001; Abkowitz et al, 2003). This recently published study confirmed that few HSC exit the marrow, transit the blood and re-enter the marrow during 3–12 weeks of parabiosis, and thus suggest that, under normal homeostatic conditions, HSC exit the marrow as a one-way process, potentially a death pathway. In contrast, however, cytokine exposure significantly increased the return of HSC to the marrow, as well as their mobilization to circulating blood (Abkowitz et al, 2003).

The proliferation of HSC is increased during mobilization, followed by subsequent migration and egress from the BM in the PB (Abkowitz et al, 2003). Mobilization of haematopoietic stem cells can be achieved by a diversity of agents (Table I) including haematopoietic growth factors, such as stem cell factor (SCF), ftl-3 ligand (Brasel et al, 1996), interleukin 3 (IL-3) (Orazi et al, 1992), granulocyte–macrophage colony stimulating factor (GM-CSF) (Socinski et al, 1988) or granulocyte colony-stimulating factor (G-CSF) (Dührsen et al, 1988), chemotherapeutic agents, combinations of cytokines and chemotherapeutic agents or certain chemokines. The use of haematopoietic growth factors like GM-CSF led to a 45-fold increase of circulating progenitor cells in mice (Brasel et al, 1997), whereas for G-CSF an increase of up to 100-fold in mice (Molineux et al, 1990; de Haan et al, 1995) was shown. Numerous other studies have not detected significant differences between the numbers of GM-CSF- and G-CSF-mobilized PBSC, as has been reviewed elsewhere (Gazitt, 2002). The combination of a mobilization chemotherapy with G-CSF leads to an additional sevenfold increase in mobilization yield compared with chemotherapy alone, whereas an additional fivefold increase in the PBSC yield was detected with the combination of chemotherapy and GM-CSF.

Table I.  Classical cytokines used for PBSC mobilization.
Primary target cellTime to
peak response (d)
Increase in
blood HSC
  • *

    World Health Organization grade III/IV side-effects.

SCFPluripotent, mast cells7–1020 xYesStiff et al (2000)
IL-3pleiotropic155xYesOrazi et al (1992)
GM-CSFmyeloid745 xYesSocinski et al (1988)
G-CSFmyeloid4–520–100 xNoDührsen et al (1988)

The objective of this review is to give an overview of the variables that influence PBSC yields. Emphasis is placed on new strategies to optimize the blood stem cell mobilization in the clinical setting, which is particularly relevant for heavily pretreated patients.

What are haematopoietic stem cells?

There are different definitions for ‘stem cells’ and ‘progenitor cells’ as well as for ‘CD34+ cells’ used in the actual literature. This causes confusion. We would support the use of the definition of Weissman (2000): stem cells are regarded as clonogenic cells capable of self-renewal and multilineage differentiation. Progenitor cells are oligo-lineage cells that are already more restricted in their differentiation potential and are not able to self-renew. The expression of the antigen CD34 and lineage negativity are often used in clinical practice as a surrogate markers for haematopoietic stem and progenitor cells.

The percentage of CD34+ cells in peripheral blood under steady-state conditions is about 0·06% and about 1·1% in bone marrow (Fruehauf et al, 1995; Körbling & Anderlini, 2001). They are thought to play an important role both in short-term engraftment, for rapid haematopoietic reconstitution in patients receiving BM ablative doses of irradiation and chemotherapy, as well as long-term, stable engraftment. Nevertheless, their characterization remains difficult, as has recently been reviewed elsewhere (Wagers et al, 2002). At least two primitive cell subsets can be defined by their immunophenotype: CD34+/CDw90(Thy1)+ (Craig et al, 1993) and CD34+/CD38 cells (Terstappen et al, 1991). Besides the well-characterized CD34+ HSC, there is evidence that CD34-negative HSC with engraftment potential and distinct HSC characteristics exist in mice as well as humans (Goodell et al, 1997).

Interestingly, corresponding to their functional differences, CD34+ cells from the BM and the PB show a differential expression of genes: BM CD34+ cells cycle more rapidly, whereas mobilized PBSC are more deeply arrested in G0 and are composed of a greater number of quiescent stem cells (Uchida et al, 1997; Fruehauf et al, 1998). Steidl et al (2002) showed that nine cell-cycle-associated genes (e.g. CDC25A, B-MYB, PLK, CDC20, UBCH10) and 11 genes required for DNA synthesis (e.g. ligase 1) were expressed at distinctly higher levels in BM compared with circulating CD34+ cells, indicating a higher cycling activity of the former. The expression of apoptosis genes, including caspases 3, 4 and 8, was higher in circulating CD34+ cells, whereas the expression of anti-apoptotic cytoplasmic antiproteinase 2 was decreased. This might serve as a physiological mechanism to prevent uncontrolled proliferation once a haematopoietic stem cell has left the BM.

Biology of pbsc mobilization

The BM provides a suitable microenvironment for stem cells. It is a reservoir for haematopoietic stem and progenitor cells (Reya et al, 2001), endothelial progenitors (Lyden et al, 2001), and possibly neuronal, muscle, liver and pancreas progenitor cells (Krause et al, 2001). It is not known whether these primitive regenerative cells are one or many cell populations. Under steady-state conditions, most of the stem cells are maintained in G0 phase of the cell cycle by interaction with the stromal cells in the BM (Cheng et al, 2000), while there is only a small proportion of stem cells in S or G2/M phase of the cell cycle. Apparently, adhesive interactions between the CD34+ haematopoietic stem cells with cellular and matrix components of the BM environment are involved in mobilization (Papayannopoulou et al, 2001). Primitive haematopoietic stem cells express a wide range of cell adhesion molecules (CAM), including members of the integrin, selectin, immunglobulin superfamily and CD44 families of adhesion molecules. BM stromal cells express the ligands of many of these CAM. The role and relative contributions of the many individual CAM-ligand pairs in the homing, lodgement and retention of primitive haematopoietic stem cells within the BM is complex and is not fully elucidated. Mobilization of haematopoietic stem cells is believed to result from cytokine-induced functional changes in the adhesion profile expressed by the haematopoietic stem cells, facilitating egress from the BM. Mobilized CD34+ cells show significant reduction of c-kit, CD18+/CD11a+, CD49d, CD62L and CXCR4 compared with steady-state BM and PB (Kollet et al, 2001; Lapidot & Petit, 2002; Petit et al, 2002). The most extensively characterized adhesion molecules are: CD49d [very late antigen (VLA)-4] and its ligand, fibronectin (Papayannopoulou et al, 2001); CD49e (VLA-5) and its receptor, vascular adhesion molecule-1 (VCAM-1); and CD62L (L-Selectin), which mediates the initial contact of leucocytes with the endothelium and is also highly expressed on circulating CD34+ cells, suggesting its role for the homing of stem cells following transplantation (Dercksen et al, 1995; Fruehauf et al, 2003). In the mouse model, it was shown that VLA-4 and VCAM-1 were involved in the retention of stem cells in the marrow, as in vivo treatment of the mice with antibodies to VLA-4 or VCAM-1 induced CD34+ mobilization (Kikuta et al, 2000; Papayannopoulou et al, 2001).

The reduced c-kit (CD117) expression before the egress of HSC in the circulation is inversely correlated with stem cell yield. Recently, it was demonstrated that the activation of the metalloproteinase (MMP)-9, leading to the release of c-kit-L, is a decisive checkpoint for the mobilization of haematopoietic stem cells as it promotes the recruitment of stem cells into peripheral blood (Heissig et al, 2002). The importance of metalloproteinases in the mobilization process has also been shown by studies on IL-8-induced mobilization. This HSC mobilization is preceded by a rise in MMP-9, leading to degradation of critical extracellular matrix components in the rhesus monkey model, and neutralizing antibodies directed against MMP-9 partially blocked IL-8-induced HSC mobilization (Pruijt et al, 1999). A higher expression level of CXCR4 is found on BM CD34+ cells compared with PB CD34+ cells (Deichmann et al, 1997). The upregulation of CXCR4 and the reduction of its ligand, stromal cell-derived factor-1 (SDF-1), is regarded as a prerequisite for mobilization (Petit et al, 2002). SDF-1 is expressed by BM stromal cells. The differential expression of this chemokine receptor supports the model that the interaction of SDF-1 and CXCR4 is crucial for the adhesion and differentiation of haematopoietic progenitors in BM (Peled et al, 1999; Fig 1). Consistent with these observations, ‘poor mobilisers’ showed significantly higher SDF-1, CXCR4 and VLA-4 expression than ‘good mobilisers’ (Gazitt & Liu, 2001; Gazitt et al, 2001) irrespective of the growth factor used for mobilization. Blockage of the G-protein-coupled CXCR4 receptor signalling by pertussis toxin led to PBSC mobilization in an experimental setting (Papayannopoulou et al, 2002).

Figure 1.

Proposed role of the CXCR4/SDF-1 chemokine receptor/ligand pair for mobilization of haematopoietic stem cells (HSC). G-CSF induces downregulation of SDF-1 in bone marrow stroma and upregulates matrix metalloproteinase-9 (MMP-9), neutrophil elastase (NE) and cathepsin G (CG). In the presence of proteases, an N-terminal cleavage of the CXCR4 receptor occurs. The CXCR2 agonist GROβ, or its derivative SB-251353, act via induction of MMP-9. The SDF-1 agonist CTCE0021 competes with stroma-bound SDF-1 for CXCR4 binding. The small molecule AMD-3100, a bicyclam, is also a partial CXCR4 agonist. The cyclic peptide TC14012 is an inverse CXCR4 agonist that does not signal and reduces PBSC mobilization. Chemokine receptor signalling occurs through the G 1 alpha (G1α) protein (Vlahakis et al, 2002). N,N-terminal CXCR4 epitope.

G-CSF is the most established agent for the mobilization of stem cells in current clinical practice (Table I). Interestingly, it appeared that G-CSF receptor is required for G-CSF-induced proliferation of HSC in the BM but not for G-CSF-induced mobilization, as has been shown in G-CSF-receptor-deficient mice (Liu et al, 2000): G-CSF-dependent signals seem to act indirectly to mobilize stem cells from the BM. In this model, the first step toward HSC mobilization is the activation of a subset of mature haematopoietic cells by the mobilization stimulus such as G-CSF; the second step is the generation of secondary signals (protease release, SDF-1 modulation) by these activated cells, which in turn leads to mobilization. Changes in the BM microenvironment might be more important for a successful stem cell mobilization than changes within the HSC. Mobilization of HSC by G-CSF coincides in vivo with the cleavage of the N-terminus of CXCR4 on HSC resident in the BM (Fig 1). The N-terminus of CXCR4 on PB-derived CD34+ HSC, isolated from G-CSF-mobilized donors, was similarly cleaved (Levesque et al, 2002). More systematic exploration of this network resulted in the assumptions that: (1) neutrophil elastase (NE), cathepsin G and MMP-9 are not required for HSC mobilization by G-CSF in mice; (2) VCAM-1 cleavage may contribute, but is not required for HSC mobilization by G-CSF; (3) G-CSF treatment induces a decrease in BM SDF-1 protein that correlates well with HSC mobilization, but this can occur in the absence of neutrophil elastase and cathepsin G (Liu et al, 2000). Overnight culture of G-CSF-primed PBSC in a protease-free environment restored the expression of the N-terminal CXCR4 epitope. Therefore, it is likely that recovery of full-length CXCR4 expression occurs on HSC in the PB, facilitating their homing into the BM of the recipient (Levesque et al, 2002). These results suggest that the CXCR4/SDF-1 pathway plays a key role in PBSC mobilization (Fig 1).

Lessons from clinical experience

The aim of PBSC mobilization is to obtain at least 2·0 × 106 CD34+ cells per kg body weight (bw) as there is a clear inverse correlation between the amount of CD34+ cells infused below this threshold and the achievement of a rapid and sustained engraftment following myeloablative treatment (Siena et al, 1991, 2000). There are numerous studies analysing predictive factors for mobilization and the characteristics of the ‘poor mobilisers’.

Patient-associated variables

Sautois et al (1999) showed that a leucocyte count of > 5 × 109/l, more than 5% myeloid progenitors in the PB and more than 0·02 × 109 CD34+ cells/l PB were associated with higher PBSC yields in patients mobilized with chemotherapy and G-CSF. The good correlation between the CD34+ cell count in peripheral blood and the yield in the PBSC collection has been shown by many groups (Haas et al, 1994; Ho et al, 1996). In a study of 41 consecutive patients, Marques et al (2000) reported that an early total white blood cell recovery was predictive of a low number of apheresis and high CD34+ cell yield.

Healthy donors have been reported to show a high interindividual variation in PBSC mobilization using G-CSF (Grigg et al, 1995; Anderlini et al, 1997; Holm, 1998; de la Rubia et al, 2002). The roles of donor age and the G-CSF administration schedule are controversial. A significant negative influence of these two variables on the CD34+ cell mobilization has been shown when the aim was to collect ≥ 4·0 × 106 CD34+/kg bw (de la Rubia et al, 2002). This could not be confirmed by others or the correlation seen was rather weak (Miflin et al, 1996; Tabilio et al, 1997). Consequently, Anderlini et al (1997) concluded that it is not possible to exclude a healthy individual as PBSC donor because of clinical characteristics associated with a low PBSC yield. In terms of tolerance and mobilization capacity of CD34+ cells, G-CSF was proven to be superior to GM-CSF in healthy donors (Fischmeister et al, 1999). The acute G-CSF-associated side affects are generally mild: bone and muscle pain, and slight spleen enlargement (Platzbecker et al, 2001). Severe allergic reactions are rare. Several reports have demonstrated that G-CSF administration induces a transient mild hypercoagulable state, therefore, caution should be applied when considering donors with a history of peripheral vascular disease, myocardial infarction or stroke (Gutierrez-Delgado & Bensinger, 2001). In conclusion, the long-term side-effects of G-CSF administration to normal donors need to be further studied (Beelen et al, 2002), and a longer follow-up using large registry databases is required (Gutierrez-Delgado & Bensinger 2001; de la Rubia et al, 2002).

The interindividual differences in PBSC mobilization may be partially due to genetic background. Mouse studies have shown that mobilization efficacy is the result of multigene traits controlled by loci on chromosomes 2 and 11 (Hasegawa et al, 2000). In general, poorly mobilizing mice strains also have low stem cell concentrations in the BM. As SDF-1 is constitutively expressed by BM stromal cells and other tissues, and has also an important role for stem cell homing (Shirozu et al, 1995), Benboubker et al (2001) investigated the relationship between SDF-1 gene polymorphism and mobilization capacity in 63 patients with malignancies. They demonstrated that the presence of the SDF-1 3′-A allele was a predictive factor for high levels of circulating CD34+ cells. As the SDF-1 genotype was not absolutely predictive for moderate or poor mobilization, interference with others factors must be considered. Little is known about the association of certain human leucocyte antigen (HLA) genotypes and mobilization potential.

We and others have observed that PB CD34+ cells and colony forming cells during steady-state haematopoiesis are measures of a patient's mobilisable pool after G-CSF and chemotherapy (Fruehauf et al, 1995, 1999; Haug et al, 1997), or after G-CSF alone (Brown et al, 1996; Husson et al, 1996; Ashihara et al, 2002). BM CD34+ cells were not predictive of the mobilization yield. Prospective studies are necessary to confirm that the expression level of molecular markers, such as CXCR4 or VLA-4 etc., really can predict poorly mobilizing patients.

Influence of the diagnosis on the pbsc yield

In stem cell disorders, the diagnosis influences the PBSC mobilization ability of patients: patients with acute myeloid leukaemia (AML) show lower CD34+ cell counts in the peripheral blood compared with other malignancies. AML patients may have limited marrow resources of CD34+ cells with a slower release of these cells into the peripheral blood during the apheresis procedure than observed in non-leukaemic individuals (Yu et al, 1999). On the other hand, Schlenk et al (1997) found that 83% of 36 patients with AML were able to collect sufficient PBSC for transplantation (target: > 2·5 × 106 CD34+ cells/kg bw) after intensive double induction with ICE (idarubicine, cytarabine, etoposide) and first consolidation with HAM (high-dose cytarabine and mitoxantrone).

The mobilization capacity of patients with haematological malignancies is, in general, lower than in patients with solid tumours. For example, patients with non-Hodgkin's lymphoma and Hodgkin's disease patients are poorer mobilisers compared with breast or testicular cancer patients, or patients with other solid tumours (D'Arena et al, 1998; Weaver et al, 1998; Gazitt et al, 1999; Sautois et al, 1999). This is most likely to be due to the more pretreatment of patients with haematolymphoid malignancies. Delayed mobilization kinetics in patients with Hodgkin's disease has also been associated with an increased risk of the development of myelodysplasia or secondary AML after transplantation (Fung et al, 2002).

Influence of age and sex of tumour patients on pbsc yield

The results regarding the influence of age on the PBSC yield are controversial, as already pointed out for the healthy donors. In children, there was no correlation between PBSC yield and the patient's age (Kanold et al, 1998; Witt et al, 2001). For patients with gynaecological cancer, an age of less than 50 years was associated with better PBSC mobilization (Kurata et al, 2000). The same holds true for the influence of sex. Some authors reported a significantly higher probability of mobilization failure in women than in men (Perea et al, 2001). Other studies could not confirm this but related the lower amount of total CD34+ cells in the leukapheresis product to the (generally) lower body weight of women compared with men. No correlation between CD34+ cell yield and sex was observed in children (Witt et al, 2001).

Is bone marrow infiltration a negative predictor for the pbsc yield?

BM involvement does not seem to be an independent factor with significant adverse influence on PBSC mobilization, although reduction of tumour infiltration within the marrow environment may improve the chances of successful mobilization (Tarella et al, 1999). In multiple myeloma, the results are contradictory: some authors report that BM involvement significantly affects adequate collections (Demirer et al, 1997; Ketterer et al, 1998) while others have failed to show any effect (Goldschmidt et al, 1997; Marit et al, 1998). These discrepancies may be related to the focal infiltration pattern of the bone marrow with plasma cells and the high likelihood of sampling errors.

Heavily pretreated patients have a lower pbsc yield

Many studies proved that prior radiotherapy and chemotherapy with alkylating agents (melphalan, carmustine) adversely affect mobilization (Watts et al, 1997; Clark & Brammer, 1998; Ketterer et al, 1998; Marit et al, 1998; Russell et al, 1998; Perea et al, 2001). Melphalan is particularly stem-cell toxic and even a low dose given orally before mobilization results in reduced CD34+ cell mobilization (Tricot et al, 1995;Prince et al, 1996; Demirer et al, 1997; Goldschmidt et al, 1997; Boccadoro et al, 2002). Fludarabine pretreatment was proven in several studies to be associated with poor PBSC mobilization (e.g. Ketterer et al, 1998) as well as other stem-cell toxic agents such as nitrogen mustard, procarbazine or more than 7·5 g cytarabine chemotherapy remobilization (Moskowitz et al, 1998). Combinations of fludarabine and cyclophosphamide were especially toxic and durably impaired steady-state G-CSF PBSC mobilization and harvest (Tournilhac et al, 2002).

Prognostic scoring indices to predict poor mobilization have been developed for certain patient groups. Perea et al (2001) found that in multiple myeloma patients a period between diagnosis and mobilization therapy of more than 12 months was an adverse factor, as well as a high BM infiltration with plasma cells and prolonged exposure to alkylating agents for a mobilization goal of 2·5 × 106 CD34+/kg bw. Corso et al (2000) aimed to achieve 0·004 × 109 CD34+/l PB, and found low white blood cell and platelet count, prior exposure to melphalan, and an interval longer than 6 months from the start of treatment to be adverse variables. In an univariate analysis of 54 Hodgkin's patients, Canales et al (2001) found poorer PBSC mobilization in those who received at least two courses of mini-BEAM, who received a high number of different chemotherapy regimens, with a chemotherapy score > 30 and who received more than 9 months of alkylating agents. Similar results were seen by another group (Sugrue et al, 2000) in 44 consecutive Hodgkin's and non-Hodgkin's lymphoma patients who underwent autologous transplantation between 1996 and 1998. Independent predictive features for poor mobilisers were a lack of significant increase in white blood cell count on the first day of apheresis and the number of pretreatment regimens. Interestingly, they found that poor mobilization predicted a worse outcome after autografting for lymphoma patients. This was not confirmed by Stockerl-Goldstein et al (2000), who did not find any difference concerning relapse, event-free survival and overall survival in 172 good and poorly mobilizing non-Hodgkin's lymphoma patients.

Although Clark and Brammer (1998) did not see a correlation between the time of previous chemotherapy and harvest outcome, several authors have shown that mobilization is negatively affected by a preceding cytotoxic treatment when the interval is short (Haas et al, 1994; Akard et al, 1996; Tarella et al, 2002). Tarella et al (1999) showed, for 39 indolent lymphoma patients, that a chemotherapy-free interval of more than 66 d was sufficient to allow adequate marrow regeneration and optimal PBSC mobilization (P = 0·0015). Such a delay is not acceptable in aggressive neoplasms with a high risk of relapse. Marrow-sparing drugs, such as dexamethasone or anti-CD20 monoclonal antibody, could be a possible therapy to achieve tumour control during this time (Maloney et al, 1994; Voso et al, 2000). We observed a trend that patients who received more than six cycles of chemotherapy (CT) premobilization reached their CD34+ cell peak in the PB later (Seggewiss et al, 2002) and showed a reduced maximal number of CD34+ cells in the PB (e.g. 5% of all patients with < 4 CT reached the CD34 peak later than d 20 vs 12% of all patients with > 6 CT). This finding is in agreement with previous studies supporting that PBSC collection should be performed early in the course of disease to avoid chemotherapy-induced stem cell damage (Goldschmidt et al, 1997; Fermand et al, 1998; Neben et al, 2002).

Controversies about the effect of the mobilization chemotherapy regimen on the pbsc yield

Cyclophosphamide (at 1–7 g/m2) is a widely used mobilization regimen. Fitoussi et al (2001) showed that cyclophosphamide at 4 g/m2 decreases haematological and extra-haematological toxicity with an equivalent CD34+ collection compared with 7 g/m2, while others found a dose–response effect of cyclophosphamide on the mobilization efficiency (Goldschmidt et al, 1997). Combined ifosfamide/epirubicin and standard dose G-CSF mobilized sufficient PBSC without serious side-effects in multiple myeloma patients, leading to a high proportion of complete remission after tandem transplantation with high-dose melphalan (Arland et al, 2001). A combination therapy, containing mitoguazone, ifosfamide, methotrexate and etoposide (MIME), followed by G-CSF administration mobilized sufficient PBSC in 130 of 141 heavily pretreated relapsed lymphoma patients (Aurlien et al, 2001). Docetaxel with significant antineoplastic activity against breast and ovarian cancer combined with G-CSF was efficiently used to mobilize PBSC with minimal toxicity in these tumour patients (Prince et al, 2000). Paclitaxel combined with cisplatin has been shown to be less effective for PBSC mobilization in patients with gynaecological cancers compared with other platinum-based chemotherapies (Kurata et al, 2000). PBSC mobilization with paclitaxel/ifosfamide and G-CSF with or without amifostine in 40 germ cell tumour patients led to higher numbers of CD34+ cells in the circulation in the amifostine group without enhancement of the overall collection efficiency (Rick et al, 2001). There are only a few studies that deal with the optimal time point of apheresis and mobilization kinetics according to the mobilization chemotherapy: Dettke et al (2001), in a report of breast cancer patients, started the PBSC collection at a median of 9 d in 47 patients receiving epirubicin/paclitaxel and G-CSF, and a median of 13 d in 39 patients who were mobilized with cyclophosphamide and G-CSF. Interestingly, the kinetics were different, with a peak at the first apheresis day and rapid decline of CD34+ cell count in the cyclophosphamide group, and a steady mobilization in the epirubicin/paclitaxel group. The authors concluded that the mobilization regimen should be taken into consideration before scheduling PBSC apheresis.

Response to cytokines versus cytokines combined with chemotherapy

Today, G-CSF and GM-CSF are the most frequently used cytokines for PBSC mobilization in patients. No difference in their mobilization ability was found for these two cytokines (Grigg et al, 1995; Hohaus et al, 1998). The response to these cytokines is dose dependent, therefore, a higher G-CSF dosage is recommended in poorly mobilizing patients (Grigg et al, 1995; Hoglund et al, 1996; Stiff, 1999). This is only true if the cytokine is applied without chemotherapy; after chemotherapy, the mobilization yield is not dependent on the G-CSF dose (Martin-Murea et al, 1998). Several studies have shown that PBSC mobilization is more effective using chemotherapy and cytokines like G-CSF, compared with cytokines alone (Bensinger et al, 1995; Demirer et al, 1997; Russell et al, 1998). It is, therefore, advisable to use a disease-specific chemotherapy combined with G-CSF or other cytokines, especially in patients with disease progression.

Management of poor mobilisers

A subset of patients, often heavily pretreated, do not achieve satisfactory PBSC mobilization. Kobbe et al (1999) suggested that poor mobilisers or heavily pretreated patients could be re-mobilized with a high dose G-CSF (12·5–50 µg/kg bw) to obtain sufficient CD34+ cells. On the other hand, after chemotherapy the mobilization yield is not dependent on the G-CSF dosage (see above), so that there is no rationale for dose escalation of G-CSF in this context. Altes et al (2000) found that a combination therapy with ifosfamide, etoposide (VP-16), cytarabine and methylprednisone (IAPVP-16) plus G-CSF is a sufficient mobilization regimen for heavily pretreated lymphoma patients and known poor mobilisers. The BM harvest is in these patients often of limited value (Stiff, 1999). Alternative approaches including new cytokines or combination of cytokines have to be considered (Tarella et al, 2002). Besides, it has been shown in mice and non-human primates that the intervals between mobilization and remobilization using a combination of G-CSF/SCF are crucial (Shi et al, 2001): an interval of 2 weeks led to a decrease in mobilization capacity whereas, after an interval of 4 weeks, an increase in the CD34+ cell yield could been seen. This was also found for humans who were regarded as ‘poor mobilisers’ and failed the first mobilization attempt (Weaver et al, 1998; Watts et al, 2000). The reasons for the improvement are not clear. It might be either as a result of the recovery time of the CD34+ pool or an increased G-CSF dosage in the second mobilization attempt.

New strategies to optimize pbsc mobilization

A multitude of agents are being developed and tested to be used alone or in combination with G-CSF or chemotherapy (Table II) in order to successfully remobilize poor mobilisers, reduce the number of apheresis sessions and/or the amount of G-CSF required for successful PBSC mobilization. Furthermore, the combinations and new agents are used to improve the haematopoietic recovery after PBSC transplantation and, last but not least, to reduce the costs of transplantation in the health care system. The most interesting compounds will be discussed in more detail below.

Table II.  New agents for PBSC mobilization.
CompoundPhase of developmentMechanism of actionTime to responseIncrease in blood HSCDosageReference
  1. hGH, human growth hormone; CT, chemotherapy; n.g., not given.

Pegylated G-CSFClinical, phase II/III 5 d20–100 x30–300 µg/kg,
single dose
Johnston et al (2000)
G-CSF + hGHClinical, phase II/III 5 dfivefold higher than G-CSF aloneG-CSF 5 µg/kg/d hGH 100 mg/kg/d Carlo-Stella et al (2002a)
G-CSF + Defibrotide (DEF)Preclinical, non-human primatesFibrinolysis5 dsix to sevenfold higher than G-CSF aloneG-CSF n.g. DEF 15 mg/kg/h Carlo-Stella et al (2002b)
CT + BB-10010Clinical, phase I/IIMIP-1α analogue, proliferation inhibition,
HSC protection
14–20 d after CTfivefold higher CFU-GM than CT alone; CD34+ increase not
100 µg/kg/d, d − 1 to + 6 after chemotherapy Clemons et al (1998)
AMD-3100Clinical, phase ICXCR4 chemokine receptor antagonist, small
molecule (bicyclam)
6–9 h7–20 x, CD34+ cells40–240 µg/kg s.c.Liles et al (2001) Dale et al (2002)
G-CSF + AMD-3100Clinical, phase I 5 d G-CSF 6 h AMD-3100threefold higher than G-CSF alone10 µg/kg G-CSF, d 1–5 160 µg/kg
AMD-3100, d 5
Liles et al (2002)
CTCE0021Preclinical, miceSDF-1 agonist, peptide4 h8 x, CFU-GM25 mg/kgPelus et al (2002a)
GROβPreclinical, miceCXCL2 agonist4–6 h20 x, CFU-GM2·5 mg/kgPelus et al (2002b)
SB-251353Preclinical, non-human
Truncated GROβ, peptide50–100 min8–10 x, CFU-GM0·5–1 mg/kgKing et al (2001)
G-CSF + SB-251353Preclinical, non-human primates 4 d G-CSF 60 min SB-251353threefold higher than G-CSF alone10 µg/kg G-CSF, d 1–4
250 µg/kg
SB-251353, d 4
King et al (2001)

Recently, a pegylated G-CSF (pegfilgrastim) has become available. The drug was approved in the USA for reduction of neutropenia following myelosuppressive therapy in patients with non-myeloid malignancies. The administration of a single dose of 30–300 µg/kg pegfilgrastim resulted in a significant mobilization of CD34+ cells in healthy donors (Molineux et al, 1999) and tumour patients treated with chemotherapy (Johnston et al, 2000). In patients with non-small-cell lung cancer (NSCLC) who received 30, 100 or 300 µg/kg pegfilgrastim in steady-state haematopoiesis, the median peak CD34+ cell count was 0·012, 0·061 and 0·043 × 109/l PB, respectively, compared with 0·005, 0·068 and 0·028 × 109/l PB if it was administered after chemotherapy. The drop at the highest dosage is probably due to an overproportionate increase in neutrophils, which take up and degrade the cytokine so that lower serum levels remain available for induction of PBSC mobilization. In comparison, G-CSF (filgrastim) 5 µg/kg/d mobilized a median of 0·015 or 0·057 × 109 CD34+ cells/l when administered prior to or after chemotherapy respectively (Johnston et al, 2000). Pegfilgrastim seems particularly attractive for PBSC mobilization in the outpatient setting because problems with patient compliance are reduced (Noga et al, 2002).

The combination of stem cell factor (SCF), a cytokine acting on early stem cells, with G-CSF resulted in improved mobilization connected with reduced apheresis numbers in multiple myeloma patients (Facon et al, 1999). In 44% of heavily pretreated patients with Hodgkin's disease or non-Hodgkin's lymphoma, it resulted in a successful mobilization (≥ 5 × 106 CD34+ cells/kg bw) compared with only 17% in the G-CSF group (Stiff et al, 2000). However, five of 102 patients experienced severe mast-cell-mediated reactions following SCF + G-CSF. Therefore, clinical development of SCF for PBSC mobilization has not been pursued further.

In poor mobilisers, the combination of G-CSF plus GM-CSF, which was already investigated by Corringham and Ho (1995), appeared to be as effective as high-dose G-CSF for PBSC mobilization, at a lower cost (Bashey et al, 2000; Stiff et al, 2002).

Growth hormone influences immune development and function. It was shown that it reversed the myelosuppressive effect of azidothymidine (Murphy et al, 1992). Therefore, it was tested in mice and shown to promote haematopoietic reconstitution following syngeneic BM transplantation (Tian et al, 1998). The mechanism by which recombinant human growth hormone (rhGH) promotes haematopoietic reconstitution is not yet clear. Based on the favourable preclinical data, Carlo-Stella et al (2002a) conducted a clinical study of human growth hormone combined with G-CSF for the treatment of poor mobilisers. Patients received two consecutive cycles of the same chemotherapy regimen, including DHAP (dexamethasone, cytarabine, cisplatin), ifosfamide–vinorelbine, VAD (vincristine, doxorubicin, dexamethasone), or FLAG-Ida (fludarabine, cytarabine, G-CSF, idarubicin). After the first cycle, PBSC were mobilized using rhG-CSF alone (5 mg/kg/d, s.c.) whereas, after the second cycle, PBSC mobilization was elicited by rhG-CSF and rhGH (100 mg/kg/d, s.c.). Compared with rhG-CSF alone, the combined rhGH/rhG-CSF treatment induced significantly higher (P < 0·005) peak values for blood CD34+ cells (0·006 ± 0·001 × 109/l vs 0·033 ± 0·009 × 109/l). No specific side-effect could be ascribed to the rhGH therapy, except a transient hyperglycaemia requiring insulin therapy in one patient with reduced glucose tolerance. Reinfusion of rhGH/rhG-CSF-mobilized CD34+ cells following myeloablative therapy resulted in a prompt haematopoietic recovery.

The fibrinolytic agent defibrotide (DEF) synergized with G-CSF for the mobilization of PBSC in non-human primates. As compared with G-CSF alone, the combined DEF/rhG-CSF treatment induced a sevenfold increase of high-proliferative potential colony forming cells and a sixfold increase of long-term culture initiating cells (Carlo-Stella et al, 2002b). Clinical studies with this combination are expected soon. Other combinations were less promising. Concomitant administration of G-CSF and recombinant human erythropoietin (rHuEpo) was inferior to G-CSF alone for PBSC mobilization, and intensive rHuEpo therapy was followed by a temporary impairment of PBSC mobilization by G-CSF (Piron & Beguin, 2002). The chemokines IL-8 (Laterveer et al, 1996) and macrophage inflammatory protein 2 (MIP-2) (Wang et al, 1997) have been tested in preclinical studies but were not sufficiently active to justify further clinical development. BB-10010 is a variant of the human form of MIP-1α, which has been shown in mice to block the entry of haematopoietic stem cells into S-phase and to increase their self-renewal capacity during recovery from cytotoxic damage. In murine studies, BB-10010 was shown to mobilize PBSC (Lord et al, 1995). In breast cancer patients receiving FAC (5-fluorouracil, doxorubicin, cyclophosphamide) chemotherapy with or without BB-10010 in a randomized study, mobilization of granulocyte–macrophage colony forming cells was enhanced by BB-10010 with an additional threefold increase over that generated by chemotherapy alone, giving a maximal 25-fold increase over pretreatment values (Clemons et al, 1998). However, the number of CD34+ cells mobilized was not significantly different between both groups. This compound has not been further evaluated for its PBSC mobilizing capacity (Hough et al, 2003).

Whereas G-CSF is a growth factor that induces stem cell proliferation within the BM and subsequent release into the PB, the small molecule AMD-3100, a bicyclam, is a novel drug candidate and a selective partial agonist of the chemokine receptor CXCR4, which is present on white blood cells (Fig 2). In a previous phase I study, it was noted that AMD-3100 induced leucocytosis (Hendrix et al, 2000). As SDF-1 and CXCR4 are known to play key regulatory roles in the trafficking and homing of human CD34+ stem cells to the BM, it was hypothesized that AMD-3100 may mobilize PBSC and further clinical trials were conducted (Liles et al, 2001). Results from a phase I study in 10 healthy volunteers demonstrated that AMD-3100 rapidly mobilizes PBSC. After injection of 80 µg/kg AMD-3100, peak increases in the number of stem cells were observed after only 6 h, with a rise of CD34+ cells from a baseline of 0·004 ± 0·001 × 109/l to 0·026 ± 0·01 × 109/l (Liles et al, 2001). The side-effects in the healthy volunteers were mild (perioral paresthesias, nausea and vomiting). In the mouse model, it was shown that AMD-3100 has synergistic effects with G-CSF and MIP-1α on stem cell mobilization (Broxmeyer et al, 2001). Recent clinical data confirm that AMD-3100 and G-CSF act synergistically, generating three times more stem cells than either agent alone (Liles et al, 2002). In addition, data confirm that PBSC mobilized by AMD-3100 have a higher repopulation potential in human–mouse xenografts than G-CSF-mobilized PBSC (Dale et al, 2002). Consistent with these functional properties, CD34+ PBSC mobilized by AMD-3100 coexpressed the primitive stem cell antigen CD90 in 50 ± 3% whereas, comparable to historical controls (Haas et al, 1995), only 23 ± 9% of G-CSF-mobilized PBSC coexpressed this marker (Liles et al, 2002). Interestingly, AMD-3100 also led to a 40–60% reduction of lymphoma growth in a xenotransplantation model (Paul et al, 2002) and is active against acute lymphoblastic leukaemia in vitro (Juarez et al, 2002). Therefore, AMD-3100 may also have an in vivo purging effect during PBSC mobilization of lymphoma patients.

Figure 2.

Chemical structure of AMD-3100, a selective partial CXCR4 agonist. The compound is a bicyclam with the chemical formula: 1,1′-[1,4-phenylenebis(methylene)]- bis-1,4,8,11-azatetradecane.

Interestingly, inhibition of the SDF-1α/CXCR4 axis blocked the G-CSF-induced mobilization of human cells in non-obese diabetic/severe combined immunodeficient mice in a different set of experiments (Petit et al, 2002). This effect was observed with either neutralizing anti-CXCR4 monoclonal antibodies (mAb), SDF-1α mAb or the cyclic peptide TC14012, a SDF-1α competitive inhibitor that binds CXCR4 but does not signal (Tamamura et al, 2002; Fujii et al, 2003). In all cases, a significant reduction of mobilized haematopoietic cells in the blood circulation was observed whereas the cellularity in the BM was similar, which suggested that the cells remained in the BM and could not egress. The different mechanisms of action of the CXCR4 agonists AMD-3100 and those of the TC14012 class (here T140) have recently been elucidated. Zhang et al (2002) identified a constitutively active mutant of CXCR4 by coupling CXCR4 to the pheromone response pathway in yeast. Conversion of Asn119 to Ser or Ala in CXCR4 conferred autonomous signalling in yeast and mammalian cells. Exposure to AMD3100 induced partial G protein activation by CXCR4 wild type and this mutant, whereas T140 decreased autonomous signalling, disclosing that T140 is an inverse agonist while AMD3100 is a partial agonist.

CTCE0021 is another agent targeting the SDF-1α/CXCR4 pathway. This peptide agonist of SDF-1α rapidly mobilizes polymorphonuclear neutrophils and haematopoietic stem and progenitor cells following a single administration, and synergizes with G-CSF for both effects. However, high doses (25 mg/kg) of this peptide are required for therapeutic efficacy. In mice, the injection of CTCE0021 resulted in dose- and time-dependent increases of up to eightfold in PB granulocyte–macrophage colony forming units (CFU-GM), erythroid burst-forming units (BFU-e) and mixed lineage CFU (CFU-GEMM) by 4 h post administration, which returned to baseline within 24 h. The addition of a single dose of CTCE0021 to a multiday G-CSF regimen resulted in a synergistic 155 ± 23-fold increase in CFU-GM/ml blood (P < 0·005) (Pelus et al, 2002a).

Other CXC receptor agonists are also being applied for mobilization. In rhesus monkeys and mice, SB-251353, a truncated form of the human CXC chemokine GROβ, was administered. Unlike IL-8, an agonist for both CXCR1 and CXCR2, this new agent is a specific CXCR2 receptor agonist (King et al, 2000). It leads to a rapid (peak within 50–100 min after administration) and transient mobilization of stem cells and neutrophils in the peripheral blood. Combined with G-CSF, it led to a threefold increase in the amount of mobilized CFU-GM compared with G-CSF alone in rhesus monkeys (King et al, 2001). In mice, the synergistic mobilization of peripheral blood haematopoietic stem and progenitor cells by G-CSF plus the CXC chemokine GROβ was found to be mediated by plasma MMP-9 (Pelus et al, 2002b). Interestingly, in the murine model, SB-251353-mobilized cells resulted in faster neutrophil and platelet recovery compared with G-CSF-mobilized HSC (King et al, 2001). A clinical phase I trial of SB-251353 is being conducted with the aim of preventing chemotherapy-induced cytopenias and increasing stem cell mobilization.


Significant progress has been made in understanding the mechanisms of PBSC mobilization. This has led to the development of new agents that are already being tested in clinical trials. AMD-3100, a small synthetic molecule and a partial CXCR4 agonist, is the most promising compound in this series. Wider recognition of the new developments in PBSC mobilization described here will enable more patients to mobilize sufficient PBSC harvests and ultimately to receive a potentially curative myeloablative high-dose therapy for their underlying malignancy.


We want to thank A. D. Ho, Heidelberg, and C. E. Dunbar, Washington DC, for their critical reading of the manuscript and helpful suggestions.