G-CSF Priming of Haploidentical Bone Marrow: Effects on Cell Yield, Collection Efficiency, and Tolerogenic Graft Composition

Introduction

Allogeneic hematopoietic stem cell transplantation (HSCT) from haploidentical donors, particularly with post-transplantation cyclophosphamide, has become a well-established method for treating malignant and non-malignant diseases in patients lacking related or matched unrelated stem cell donors. Although the original protocol utilized bone marrow as the stem cell source [1], peripheral blood stem cells are now predominantly used [2]. This preference has likely arisen because granulocyte colony-stimulating factor (G-CSF)-primed peripheral blood stem cells (i.e., G-PB) provide easier collection and faster engraftment [3], a lower cumulative incidence of relapse, and superior progression-free survival, although with higher incidences of acute and chronic graft-versus-host disease (GVHD) [4]. Thus far, studies have compared G-PB with steady-state bone marrow grafts (SS-BM), rather than G-CSF-primed bone marrow grafts (G-BM).

Previous research has shown that G-CSF priming also confers beneficial effects on bone marrow grafts. In the autologous setting, its use resulted in recovery comparable to that achieved with G-PB [5]. Some small studies have examined bone marrow priming in the allogeneic setting [6,7]. G-BM can induce more rapid recovery compared with SS-BM [8,9], along with recovery similar in speed to that of G-PB [10]. A large prospective randomized study demonstrated that transplantation with G-BM resulted in engraftment comparable to that of G-PB, along with significantly lower rates of GVHD [11]. One prospective study assessed the influence of G-CSF priming of bone marrow on transplant outcomes, including GVHD incidence, relapse, and overall survival. That study revealed a significantly lower incidence of acute GVHD in the primed group, without significant differences in chronic GVHD, relapse rates, or overall survival [12].

Only a few studies in the haploidentical HSCT setting have evaluated G-BM as the sole cell source. These reports indicated rapid engraftment and reduced acute GVHD compared with historical data [9,13], as well as lower-than-expected rates of both acute and chronic GVHD [14]. However, none of these studies used post-transplantation cyclophosphamide for GVHD prophylaxis.

One possible explanation for the apparent lower incidence of GVHD when using G-BM is its previously demonstrated beneficial effect on tolerance induction, mediated through mechanisms such as decreased expression of adhesion molecules (very late antigen-4 [VLA-4], intercellular adhesion molecule-1 [ICAM-1], lymphocyte selectin [L-selectin], and lymphocyte function-associated antigen-1 [LFA-1]) and increased percentages of interleukin-4 (IL-4)-positive cells in naïve CD4+ and CD8+ T-cell subsets [15]. Mechanisms also include reduced secretion of interferon-γ and IL-4, along with decreased numbers of dendritic cells (DC1 and DC2) [16]. Collectively, these changes promote polarization of bone marrow naïve CD4+ and CD8+ T cells from a Th1 to a Th2 phenotype and indirectly induce T-cell hyporesponsiveness. Additionally, G-BM has been shown to contain a higher percentage of regulatory T cells (Tregs), greater proportions of CD45RA+ naïve Treg cells, higher CD69 expression on Tregs, and a lower ratio of conventional T cells to Tregs compared with G-PB grafts [17,18]. Furthermore, Franzke et al. suggested that G-CSF directly modulates T-cell immune responses through its receptor on T cells by suppressing gene expression of the interferon-stimulated gene factor 3 (ISGF3)-γ subunit/p48 in CD4+ donor T cells [19].

This evidence has generated increasing interest and raised the question of whether primed bone marrow represents an advantageous cell source [20,21]. Given the relative scarcity of data in the haploidentical transplant setting using post-transplantation cyclophosphamide prophylaxis, we aimed to examine clinical outcomes after HSCT from haploidentical donors with G-BM grafts and post-transplantation cyclophosphamide for GVHD prophylaxis, assess the potential benefits of this approach, and characterize the specific cell subsets influenced by priming in comparison with SS-BM.

Material and Methods

BONE MARROW DONORS AND HARVEST:

Bone marrow donors in the G-BM group were first-degree relatives of the recipients. Donors received 3 daily subcutaneous G-CSF injections on days −2, −1, and 0 (10 μg/kg body weight [BW]). The schedule was selected based on previously published data indicating that a dose of 10 μg/kg BW achieved earlier peak nucleated cell numbers relative to 5 μg/kg BW, without significant differences in colony-forming units per milliliter between time points [22].

Bone marrow was collected from the posterior iliac crests in accordance with institutional standard operating procedures. The harvest aimed to collect 5×10⁸/L. No adverse events related to G-CSF stimulation were recorded. Control donors were either HLA-matched first-degree relatives or unrelated donors. They did not receive stimulation prior to harvest (i.e., SS-BM). A bone marrow sample was obtained at the beginning of harvest and immediately sent to the laboratory for further analysis.

GVHD PROPHYLAXIS AND SUPPORTIVE CARE:

GVHD prophylaxis included post-transplantation cyclophosphamide administered on days +3 and +4, along with cyclosporin A and mycophenolate mofetil, as previously described [23]. Supportive care was provided according to institutional protocols. All patients received G-CSF (5 μg/kg BW) from day +5 until neutrophil engraftment.

The day of neutrophil engraftment was defined as the first day on which absolute neutrophil count (ANC) greater than 0.5×109/L was achieved; the day of platelet engraftment was defined as the first day on which a platelet count above 20×109/L was reached without platelet transfusion. Primary graft failure was defined as failure to achieve an ANC of at least 0.5×109/L by day +28, with associated pancytopenia and absent donor chimerism. Secondary graft failure was defined as ANC below 0.5×109/L after initial engraftment, with donor chimerism less than 5%, not due to relapse [24]. GVHD was graded in accordance with established criteria [25].

FLOW CYTOMETRY EVALUATION OF GRAFT COMPOSITION:

Flow cytometry analysis of unseparated cells from fresh samples was performed by staining cells with fluorochrome-conjugated antibodies. The FACS Lyric (BD Biosciences, San Jose, CA, USA) was used for acquisition, and data were analyzed with FACSuite ver1.2 software (BD Biosciences). Enumeration of CD34-positive cells followed the International Society of Hematotherapy and Graft Engineering (ISHAGE) dual-platform protocol [26]. The absolute count (cells/μL) of all cell subpopulations was calculated using leukocyte counts from the hematology analyzer Advia 2120i (Siemens Healthcare GmbH, Erlangen, Germany). Lymphocyte subpopulations were expressed as percentages of the total lymphocyte count.

STATISTICAL ANALYSIS:

Descriptive statistics were used to characterize donors, recipients, and bone marrow grafts. Survival curves were generated using Kaplan-Meier estimates. Cumulative incidence analyses, adjusted for competing risks, were conducted with R software. Graft composition was analyzed using Prism 7.0 software. Statistical significance was determined by the Mann-Whitney test. P-values <0.05 were considered statistically significant.

Results

TRANSPLANT CHARACTERISTICS AND ENGRAFTMENT:

Between May 2012 and December 2021, 61 patients underwent bone marrow stem cell transplantation from haploidentical donors. All patients were treated for hematologic malignancies, predominantly acute myeloid leukemia (n=25, 41%) and Hodgkin lymphoma (n=20, 33%); 2 patients (3%) were treated for aplastic anemia/bone marrow failure syndrome. Fifty-four patients (88.5%) received non-myeloablative conditioning (fludarabine, cyclophosphamide, and total body irradiation [FluCyTBI]), 5 patients (8.2%) received myeloablative conditioning (busulfan and cyclophosphamide [BuCy]), and 2 patients (3.3%) received myeloablative sequential conditioning (thiotepa, etoposide, cyclophosphamide – reduced-intensity conditioning [TEC RIC]). Patient and donor characteristics are summarized in Table 1.

Recipients received a median of 5.04×108/kg (range, 1.76–12.93) total nucleated cells (TNCs), 1.65×106/kg (range, 0.28–4.47) CD34+ cells, and 2.81×107/kg (range, 0.37–6.65) CD3+ cells. Engraftment was achieved in 56 patients (92%), with a median time to neutrophil recovery (ANC >0.5×109/L) of 23 days (range, 12–38) and a median time to platelet recovery (platelet count >20×109/L, independent of platelet transfusion) of 33 days (range, 15–81). Four patients (6%) experienced graft rejection, including 2 patients (3%) with primary rejection and 2 patients (3%) with secondary rejection. Three patients (5%) died of infectious complications without myeloid recovery (Table 2).

ACUTE AND CHRONIC GVHD INCIDENCE, SURVIVAL, AND MORTALITY:

Over a median follow-up interval of 40 months (range, 6–90), outcomes were favorable; the 2-year overall survival was 66% (95% confidence interval [CI], 55–79) (Figure 1A). Relapse occurred in 29% (95% CI, 18–41) of patients, whereas non-relapse mortality was 18% (Figure 1B, 1C). Acute GVHD (grade II–IV) developed in 20% of patients (95% CI, 11–30), and severe cases were uncommon (grade III–IV in 3 patients). Chronic GVHD was rare, with a 2-year incidence of 4% (95% CI, 1–11) (Figure 1D). These findings indicate a low burden of GVHD, as well as an encouraging balance between disease control and treatment-related toxicity.

EFFECT OF BONE MARROW PRIMING ON MARROW CONTENTS AND GRAFT CHARACTERISTICS:

To further investigate the changes in graft composition associated with G-CSF priming, 17 grafts from G-CSF-primed donors and 9 grafts from non-primed donors (controls) were prospectively collected and analyzed. In addition to TNC and total white blood cell contents, we assessed dendritic cell (DC) content, lymphocyte subpopulations, natural killer (NK) cell populations, and CD34⁺ cell content by flow cytometry.

Priming bone marrow donors with G-CSF resulted in significantly higher TNC yields relative to unprimed donors (6.19×108 TNCs/kg BW vs 1.5×108 TNCs/kg BW, P<0.0001), a higher concentration of mononuclear cells in the graft (26.5×109/L vs 7.2×109/L, P<0.0001), and an increased CD34+ cell concentration (2.94×108/L vs 0.87×108/L, P=0.004) (Figure 2A–2C). Priming also increased both CD4+CD3+ and CD8+CD3+ cell concentrations (5.83×108/L vs 3.76×108/L, P=0.011, and 5.23×108/L vs 3.52×108/L, P=0.044, respectively) (Figure 2D, 2E). The concentration of Tregs (CD3+CD4+CD127dimCD25+) was significantly higher in the primed group than in the unprimed group (0.5×108/L vs 0.24×108/L, P=0.0084) (Figure 2F).

The NK cell compartment was also affected. The percentage of CD56+ NK cells among lymphocytes was reduced in the primed group compared with the unprimed group (4.6% vs 12%, P=0.0007), and the percentage of CD56dim cells was significantly lower (3.2% vs 9.6%, P=0.0002) (Figure 3A, 3B). In contrast, the proportion of CD56bright cells did not significantly differ between groups (0.6% vs 0.9%, P=ns) (Figure 3C). Furthermore, priming increased all examined DC subsets. Plasmacytoid DCs were significantly more abundant in the primed group (59.1/mL vs 17.3/mL, P=0.0005), as were myeloid DC1 (15.4/mL vs 8.5/mL, P=0.025) and myeloid DC2 (3.3/mL vs 1.4/mL, P=0.04) (Figure 4A–4C).

EFFECT OF PRIMING ON THE SPEED OF COLLECTION:

To determine whether priming influences graft collection speed and reduces operative time, we analyzed the total collection time per graft in the cohort of donors whose graft composition was evaluated (17 primed donors and 9 controls). No significant difference was observed in extraction speed, calculated as the median milliliters of graft per minute (25.6 mL/min in the primed group vs 20 mL/min in the control group), or in the median time required to obtain the graft (54 min in the primed group vs 67 min in the control group, P=0.0675). However, because the collection goal was higher for the primed group (5×108 TNCs/kg BW) than for the control group (3×108 TNCs/kg BW), a significant difference was noted in the time needed to reach the benchmark of 100×108 TNCs. The primed group achieved this threshold after a median of 11.8 min, compared with 62.4 min in the control group (P<0.0001) (Figure 5A, 5B).

Discussion

The use of G-BM as a graft source for HSCT is increasingly attractive because it appears to offer advantages extending beyond higher cell yields. Knowledge regarding the effects of bone marrow priming in the haploidentical setting remains limited, given that it is largely derived from retrospective studies with relatively small patient cohorts. Our findings support and expand upon these data, demonstrating that G-BM not only provides grafts with higher cell yields but also enhances collection efficiency, shortens operative time, and modulates the graft’s immunological profile, potentially reducing the incidence of GVHD.

We observed that G-CSF priming significantly increased TNC and CD34+ cell yields relative to SS-BM, consistent with previous findings that G-BM can achieve faster engraftment than SS-BM and at rates comparable to G-PB grafts [8,11,27]. We also demonstrated that priming increases the proportion of CD34+ cells in the graft compared with unprimed controls. In our clinical cohort, this increased proportion resulted in satisfactory engraftment, consistent with previously reported findings [7,9,11,12] and reviews [20]. In addition to the effect on engraftment, the increased yields had a practical impact: graft collection times were shorter, translating into reduced time in the operating room for the bone marrow extraction team, less time under general anesthesia for donors, and fewer bone punctures. Thus, donor safety was enhanced. G-CSF administration in healthy donors has been extensively validated in terms of safety [28–30]. Considering these previous reports and our current findings of higher efficiency and safety, we propose regarding G-BM as a viable and advantageous graft source for haploidentical transplants, as well as matched related or unrelated donor settings [21].

Our cohort demonstrated low rates of acute GVHD (20%) and chronic GVHD (4%), despite the inclusion of patients (16%) undergoing a second transplantation. Although our study lacked a clinical control arm, prior work comparing G-BM with G-PB in related donor transplants revealed similar engraftment outcomes and reduced GVHD incidence with G-BM [11]. Acute GVHD rates in that study were comparable to ours, whereas our chronic GVHD incidence was notably lower, likely reflecting the use of post-transplantation cyclophosphamide [23]. A possible explanation for this lower-than-expected GVHD incidence is the higher number of Tregs observed in G-BM compared with controls. This finding is consistent with a study comparing G-BM and SS-BM, which showed an enhanced suppressive phenotype in G-BM (e.g., increased Treg and Th2 subsets) [18]. The beneficial immunomodulatory effect of bone marrow priming has also been demonstrated in comparisons with peripheral stem cell collections [17], supporting the conclusion that priming alters graft composition and induces T-cell hyporesponsiveness in bone marrow.

Another factor that may contribute to improved protection against GVHD is the difference in NK cell expansion, although conflicting data persist regarding the influence of G-CSF priming of bone marrow on NK cell populations and GVHD incidence. Our results showed that primed bone marrow contained a lower percentage of NK cells, particularly fewer CD56dim NK cells, whereas CD56bright cells were unchanged. This finding is consistent with the work of Yu et al., who reported that in vivo G-CSF administration reduced the percentage of NK cells. However, their study also demonstrated modulation of NK subpopulations, leading to an increased ratio of CD56bright to CD56dim subsets, which we did not observe [31]. The impact of NK subpopulations on GVHD incidence is not yet fully understood. Zhao et al. reported that higher numbers of CD56bright cells may predict an increased incidence of GVHD [32]. In contrast, during a study of peripheral blood stem cell grafts, Yamasaki et al. [33] found that the CD56+ cell dose was significantly and inversely correlated with GVHD incidence. Furthermore, multivariate analysis in the HLA-identical sibling setting demonstrated that a lower NK cell dose in the graft increased the risk of chronic GVHD [34]. In the haploidentical setting, a study involving SS-BM indicated that severe acute GVHD incidence was associated – among other factors – with a low ratio of CD56+ cells in the graft [35]. The effects of NK cell numbers and subsets likely differ according to the clinical setting, graft type, or donor, but another effect of priming should not be overlooked: priming influences not only the ratios of NK cell subsets but also their functional properties. Evidence indicates that CD56+/CD3− NK cells from G-CSF-mobilized products exhibit reduced cytotoxicity and proliferation [36], which may explain some of the conflicting data. Disparities in reported findings may also reflect differences in priming protocols, given that doses and durations of stimulation varied across studies: 4 to 10 mg/kg BW of G-CSF were administered over 2 to 7 days [8,11,14,27].

Plasmacytoid DCs, in particular, have been shown to exert tolerogenic effects that may contribute to GVHD mitigation under specific conditions [37]. The presence of increased concentrations of plasmacytoid and myeloid DCs in G-BM compared with controls supports the hypothesis that G-CSF priming promotes a more immunoregulatory graft profile.

Limitations of our study include its retrospective design and the absence of a control group for clinical outcomes, which prevents definitive conclusions regarding the ability of priming to reduce GVHD incidence in the haploidentical setting with post-transplantation cyclophosphamide prophylaxis. Nonetheless, our prospective analysis of graft composition provides a mechanistic basis for the favorable clinical outcomes observed and establishes a foundation for future prospective randomized studies to determine whether G-BM can achieve engraftment rates comparable to those of G-PB while avoiding the associated increase in GVHD incidence.

Conclusions

This study demonstrates that G-CSF priming enhances bone marrow collection efficiency and modifies graft composition by increasing Treg and DC contents, changes consistent with a tolerogenic immune profile. These findings are novel in the context of haploidentical transplantation and suggest a mechanism for reducing GVHD risk while maintaining engraftment efficacy. However, the clinical component of our analysis was retrospective and lacked a control group, limiting the strength of causal inferences. Prospective, controlled studies are warranted to clarify the relationship between these immunological changes and clinical outcomes, determine optimal priming regimens, and evaluate their applicability across diverse donor and conditioning settings.

Data Availability

The datasets generated and/or analyzed during the current study are available from the corresponding author upon request.