Background
HLA-haploidentical hematopoietic stem cell transplantation (HSCT) is an effective therapeutic option for patients with high-risk leukemia, and without human leukocyte antigen (HLA)-matched donors [
1]. Historically, clinical success, i.e., full donor-type engraftment in 95% of patients with acute leukemia and negligible incidence of acute and chronic graft-versus-host disease (GVHD), has been achieved with T-cell depleted (TCD) grafts containing a mega-dose of positively selected CD34
+ cells, without the use of any post-transplant immunosuppression [
2].
Granulocyte colony-stimulating factor (G-CSF) is widely employed as mobilizing agent in healthy donors and cancer patients. However, G-CSF-based regimens are associated with a 5-30% failure rate [
3]. The bicyclam AMD3100, also known as plerixafor, was approved in 2008 for use in combination with G-CSF to mobilize hematopoietic stem cells (HSC) for autologous HSCT [
4]. Plerixafor (Mozobil®, MZ) specifically and reversibly blocks the binding of C-X-C chemokine receptor 4 (CXCR4) to its natural ligand, stromal cell-derived factor 1 (SDF1), a CXC chemokine and key regulator of HSC homing and retention in the bone marrow (BM). We previously showed that G-CSF-mobilized peripheral blood CD34
+ cells retain surface CXCR4 [
5], implying that BM microenvironment might easily accommodate immigrating progenitor cells that express high levels of CXCR4 following G-CSF mobilization or stress conditions. MZ synergizes with G-CSF through its different mechanism of action, as suggested by randomized phase III studies, where plerixafor and G-CSF were shown to be superior to G-CSF alone for CD34
+ HSC mobilization and collection [
6],[
7].
Dendritic cells (DCs) are professional antigen-presenting cells triggering primary adaptive immune responses through the activation of
naïve CD4
+ and CD8
+ T cells [
8]. Initially, human DCs were categorized into type 1 (DC1) and type 2 DCs (DC2), which are functionally distinguished by pattern of cytokine production and T-cell driving capacity. Recently, 3 cell types assigned to the DC lineage have been characterized in human blood, i.e., type 1 myeloid DCs (MDC1), type 2 myeloid DCs (MDC2) and plasmacytoid DCs [
9]-[
11]. Blood CD1c
+ MDC1 efficiently cross-present soluble antigens and prime cytotoxic T cells [
12]. Human BDCA-3
+ MDC2 share some characteristics with murine CD8α
+ DCs, such as production of high amounts of IL-12p70 and interferon (IFN)-λ [
10],[
11]. By contrast, human plasmacytoid DCs secrete IFN-α and activate natural killer (NK) cells, macrophages and myeloid DCs to mount immune responses against microbial products.
There is growing evidence that the biological activities of G-CSF are not limited only to the myeloid lineage, but extend to other cell types mediating, amongst the others, inflammation, immunity and angiogenesis [
13],[
14]. Initial studies in mice supported a role for G-CSF in immune skewing towards T helper type 2 (Th2) cytokine production [
15]. In humans, G-CSF increases IL-4 release and decreases IFN-γ secretion [
16], and promotes the differentiation of transforming growth factor-β1/IL-10-producing type 1 regulatory T cells (Treg), which are endowed with the ability to suppress T-cell proliferation in a cytokine-dependent manner [
17],[
18]. Finally, G-CSF indirectly modulates DC function, by inducing hepatocyte growth factor, IL-10 and IFN-α, and mobilizes DC2 [
19]-[
21].
Currently, the use of MZ in healthy donors is off-label, with anecdotal reports describing its ‘just-in-time’ application either as single agent or after mobilization failure with G-CSF [
22]-[
24]. The few available data on immunological effects of MZ are mostly limited to cancer patients and show that CD8
+ T-cell release of IFN-γ and TNF-α may be higher in autologous grafts collected after G-CSF and MZ, compared with G-CSF alone [
25].
We recently developed a novel graft manipulation strategy aimed at extensively removing T-cell receptor (TCR)-αβ
+ T cells and CD19
+ B cells from haploidentical HSCs, prior to their infusion into children with non-malignant disorders [
26]. TCR-αβ and B-cell depletion is intended to prevent GVHD and post-transplantation lymphoproliferative disorders, respectively. The present study was designed and conducted to investigate whether and to what extent the administration of MZ, an ‘immediate salvage’ strategy in donors with suboptimal CD34-cell counts after standard-dose G-CSF, affects the cellular composition of the graft in the setting of TCR-αβ/CD19-depleted haploidentical HSCT for children with hematological disorders.
Discussion
Herein, we show that the addition of ‘immediate salvage’ MZ to a standard, G-CSF-based mobilization regimen augments HSC yield in HLA-haploidentical donors showing less-than-optimal CD34-cell mobilization, and that MZ administration affects the graft cell composition. Thirty-two percent of our donors were operationally defined as PMs and were given MZ to increase HSC mobilization. More than 95% of the donors collected the required mega-dose of HSCs with a single apheresis session.
Currently, anecdotal reports describe the ‘just-in-time’ application of MZ to healthy donors, either as single agent or after mobilization failure with G-CSF [
22]-[
24]. In our donor cohort, single-dose MZ given on day +5 increased the count of CD34
+ HSCs by 8.2-fold (range 1.4-29.2), compared with that measured after 4 days of G-CSF treatment. This is remarkably similar to the 8-fold increase of CD34
+ HSCs reported in donors given MZ only [
23].
The few available data on immunological effects of MZ are mostly limited to cancer patients and show that CD8
+ T-cell release of IFN-γ and TNF-α may be higher in autologous grafts collected after G-CSF and MZ, compared with G-CSF alone [
25]. We previously showed that G-CSF polarizes human T-cell and DC function towards a tolerogenic profile, implying that G-CSF-mobilized cell therapy products may be intrinsically less capable of inducing uncontrollable GVHD [
17],[
19],[
41]. This is reinforced by intriguing observations in major histocompatibility complex (MHC)-matched HSCT, where mice given G-CSF-mobilized splenocytes experienced lower rates of skin GVHD compared with recipients of MZ-mobilized splenocytes [
42]. In our healthy donors treated with G-CSF, the down-regulation of CD4
+ T-cell production of IFN-γ was not potentiated by single-dose MZ. Furthermore, IL-17 and IL-4 release by CD4
+ T cells were not affected by G-CSF and/or MZ administration. These observations are in line with pre-clinical data showing that MZ alone, in contrast to G-CSF, is unable to alter the phenotype and cytokine polarization of T cells, as well as T-cell’s ability to induce acute GVHD [
43]. It must be emphasized that, in our study, neither G-CSF nor MZ significantly impaired IFN-γ production by CD8
+ T cells. Notably, studies in mice suggest that G-CSF may separate GVHD and graft-versus-leukemia (GVL) responses by exerting suppressive effects on CD4
+ T cells, that are implicated in GVHD, while preserving the cytolytic pathways of CD8
+ T cells that are critical for effective GVL [
44]. At variance with a recent report on the immunological effects of single-dose MZ in healthy donors [
42], we were unable to detect any difference in the frequency of naïve B cells after MZ administration. Conceivably, any MZ effect on the recirculation of B-cell subsets may have been obscured by treatment with G-CSF during the 4 days preceding MZ administration. However, neither that study [
42] nor our own report identified any modification of CD4 and CD8 T-cell frequencies that could be directly attributable to MZ.
In our cohort of 90 donors, HSC mobilization with G-CSF translated into lowered frequencies of both NK cells and NK-like CD56
+ T cells, a phenomenon that was mainly accounted for by a reduction of fully mature CD56
+CD16
+ and tissue-resident CD56
+CD16
- NK cells, with preserved frequencies of immature CD56
-CD16
+ NK cells. Interestingly, the frequency of both NK cells and NK-like CD56
+ T cells was reduced in the PM group receiving single-dose MZ. Although the number of NK cells collected was not significantly different when comparing donors given G-CSF alone with those receiving G-CSF in combination with MZ, higher numbers of both NK cells and NK-like CD56
+ T cells were infused in children transplanted with G-CSF-mobilized HSC products. A potential clinical implication of this finding pertains to the field of graft engineering, insofar donor mobilization with G-CSF alone might offer an advantage over the use of G-CSF + MZ for patients with NK-susceptible hematological malignancies [
37].
As previously published, G-CSF-mobilized monocytes are functionally defective [
45]. In our study, treatment with G-CSF and MZ potently mobilized donor monocytes, especially the CD16
+ subset of intermediate/non-conventional monocytes. Although the frequency of both conventional and CD16
+ monocytes was higher in TCR-αβ/CD19-depleted HSC grafts compared with normal BM samples, the overall number of conventional and CD16
+ monocytes infused in our haploidentical HSCT recipients was not correlated with the mobilization regimen used in the donor. It has been demonstrated that macrophages generated from CD16
+ monocytes manifest higher phagocytic activity compared with macrophages derived from classical monocytes [
46]. In addition, CD14
dimCD16
+ monocytes are endowed with a unique patrolling function, as they detect virally infected and damaged cells and produce pro-inflammatory cytokines [
47]. In light of these findings, it is conceivable that CD16
+ monocytes infused with the TCR-αβ/CD19-depleted haploidentical grafts may protect the recipient from infectious episodes, while contributing to prevention of GVHD [
48].
There is also evidence that G-CSF mobilizes IL-12/TNF-α-producing, pro-inflammatory Slan-DCs [
49]. Thus, Slan-DCs may incite GVHD on the one side, while preserving GVL reactivity on the other side. In our study, the addition of MZ to G-CSF did not affect the mobilization of Slan-DCs. In addition, the frequency of Slan-DCs was lower in HSC grafts collected after the combined treatment with G-CSF and MZ. Because G-CSF administration is associated with
in vivo cleavage of the N-terminus of CXCR4 on BM-resident HSCs [
35], it is tempting to speculate that Slan-DCs, at variance with CD34
+ HSCs and monocytes [
5],[
50], may not entirely depend upon the CXCR4/SDF-1α axis for re-circulation and homing into lymphoid organs and/or tissues. Our contention is backed by experiments with CD184-12G5 antibodies showing that CXCR4 levels are preserved on the surface of Slan-DCs, but not other leukocyte subsets, analyzed after the
in vivo administration of G-CSF, when compared to G-CSF plus MZ, this likely favoring Slan-DC retention into tissues. The CD184-12G5 mAbs recognize an epitope involving the first and second extracellular domains of CXCR4, and inhibit MZ binding to CXCR4. A different anti-CXCR4 mAb, termed 1D9, binds to the N terminus of CXCR4 and is not affected by MZ [
51]. In a phase 1/2 study of chemo-sensitization with MZ in relapsed or refractory acute myeloid leukemia, a decrease in CD184-12G5 binding was observed from pre-treatment to 6 hours, followed by an increase from 6 to 24 hours, indicating CXCR4 blockade by MZ
in vivo[
52]. By contrast, labeling with CD184-1D9 after MZ treatment revealed an increased expression of CXCR4 between pretreatment and 6 hours, that remained elevated at 24 hours.
Finally, the frequency of CD1c
+ MDC1, CD141
+ MDC2 and plasmacytoid DCs [
33] was unchanged in PB of donors treated with G-CSF, either alone or in combination with MZ. This is in agreement with previous reports showing no changes in the frequency of BDCA-2
+ DCs in donors given G-CSF compared with baseline [
36]. The TCR-αβ/CD19-depleted haploidentical grafts collected after the administration of G-CSF and MZ were highly enriched with MDC1, MDC2 and plasmacytoid DCs, when compared with normal BM samples. The role played by DCs in the regulation of human GVHD and GVL responses is the subject of intense investigation. Studies of BM transplantation have shown that high numbers of plasmacytoid DCs in the graft correlate with decreased chronic GVHD, at the expense of an increased incidence of leukemia relapse [
53]. Conversely, the number of DCs in PB allografts may not predict DC reconstitution kinetics after transplantation or clinical outcome [
54]. Importantly, low DC counts at time of engraftment have been associated with worse survival, increased incidence of relapse and higher incidence of grade II-IV acute GVHD [
54].
Thus far, we have transplanted 23 children with non-malignant disorders using TCR-αβ/CD19-depleted HSC grafts [
26]. Primary graft failure occurred in 4 patients, with 3 patients developing skin-only grade 1 to 2 acute GVHD and no patient suffering from chronic GVHD. The cumulative incidence of transplantation-related mortality was 9.3%. With a median follow-up of 18 months, 21 of 23 children are alive and disease-free, the 2-year probability of disease-free survival being 91.1% [
26]. It remains to be determined whether and to what extent the DC content of our TCR-αβ/CD19-depleted HSC grafts and, in particular, the remarkably high numbers of MDC1, MDC2 and plasmacytoid DCs infused correlate with infection control, GVHD and/or leukemia recurrence.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SR, PF and VB performed the laboratory work for this study and analyzed the data; GLP, LA, EG and GC manipulated the haploidentical grafts; SC collected hematopoietic stem cells; LPB, BL and MGC cared for HSC donors; AB cared for patients; TC provided intellectual input; LM and FL provided intellectual input, analyzed the data and drafted the manuscript; SR conceived the study, participated in its design and coordination and drafted the manuscript. All authors read and approved the final manuscript.