Stem cell collection in unmanipulated HLA-haploidentical/mismatched related transplantation with combined granulocyte-colony stimulating factor-mobilised blood and bone marrow for patients with haematologic malignancies: the impact of donor characteristics and procedural settings

Unmanipulated haploidentical/mismatched related transplantation with combined granulocyte-colony stimulating factor-mobilised peripheral blood stem cells (G-PBSCs) and granulocyte-colony stimulating factor-mobilised bone marrow (G-BM) has been developed as an alternative transplantation strategy for patients with haematologic malignancies. However, little information is available about the factors predicting the outcome of peripheral blood stem cell (PBSC) collection and bone marrow (BM) harvest in this transplantation. The effects of donor characteristics and procedure factors on CD34+ cell yield were investigated. A total of 104 related healthy donors received granulocyte-colony stimulating factor (G-CSF) followed by PBSC collection and BM harvest. Male donors had significantly higher yields compared with female donors. In multiple regression analysis for peripheral blood collection, age and flow rate were negatively correlated with cell yield, whereas body mass index, pre-aphaeresis white blood cell (WBC) and circulating immature cell (CIC) counts were positively correlated with cell yields. For BM harvest, age was negatively correlated with cell yields, whereas pre-BM collection CIC counts were positively correlated with cell yield. All donors achieved the final product of ≥6 ×106 kg−1 recipient body weight. This transplantation strategy has been shown to be a feasible approach with acceptable outcomes in stem cell collection for patients who received HLA-haploidentical/mismatched transplantation with combined G-PBSCs and G-BM. In donors with multiple high-risk characteristics for poor aphaeresis CD34+ cell yield, BM was an alternative source.

Although the number of mismatched allogeneic transplants has increased steadily over the past few decades, the unmanipulated HLA-haploidentical/mismatched HSCT from related donors was still a small portion. Identifying factors associated with better quality and greater quantity of stem cells may provide an opportunity to enhance patient recovery while ensuring a safe process for stem cell donation for the HLAhaploidentical/mismatched related donors. In order to gain insight into the factors that may affect the quantity of stem cell collection for leukaemic patients who undergo unmanipulated HLA-haploidentical/mismatched related HSCT with combined G-BM and G-PBSCs, we retrospectively analysed the effects of donor characteristics and procedure factors on CD34 + cell yield from a single centre in China.

Donors
Between 2005 and December 2008, a total of 104 healthy related volunteers underwent peripheral blood stem cell (PBSC) and BM harvest in our institution. Donor assessments were in accordance with the standards of donor evaluation for allogeneic transplantations (Sierra et al., 1997). Informed consent was obtained from each donor. HLA-A, HLA-B and HLA-DR typing were performed by serology for all donor-recipient pairs. The degree of the locus mismatch conforms to the China consensus (Huang et al., 2004) on immunogenetic donor search for transplantation full mismatch. Tables 1 and 2 show the characteristics and procedural variables in the cohort, respectively.

PBSC mobilisation
The mobilisation of PBSCs was done according to our previous report (Chen et al., 2009). Each donor received the G-CSF subcutaneously at a dose of 10 μg kg −1 body weight (BW) per day (given as a split dose twice daily) at approximately the same time every day. To avoid calculation errors, the actual dose was calculated at the time of the donor health examination.

PBSC and BM collection
Leukaphaeresis was carried out on the fourth day after G-CSF mobilisation with the aid of an automated continuous flow blood cell separator (CS3000; Baxter, Deerfield, IL, USA) at a planned flow rate of 40-60 mL min −1 . The actual flow rate was calculated by dividing the total processed blood volume (mL) by the processing time (min). Aphaeresis was performed once for all cases. The donor BM was harvested at room temperature in the operation room on the fifth day after G-CSF treatment under epidural anaesthesia. The plasma and red blood cell (RBC) were excluded from the collected BM by hypothermic centrifugal machine and hydroxyethyl starch for the ABO minor and major mismatched donors, respectively. Both the plasma and RBC were separated for both the ABO minor and major mismatched donors. G-PBSCs and G-BM were transfused intravenously to the recipients just after the completion of the collection on days 1 and 2 of transplantation, respectively. Calcium was intravenously administered prophylactically during leukaphaeresis to reduce signs of hypocalcaemia. Daily counts and serum chemistry were obtained during G-CSF administration and in addition 30 days after the last administration of G-CSF. Donors complaining of bone pain received naproxen. The autologous RBC was  Unless noted otherwise, data are presented as mean (range); BM, bone marrow; BW, body weight; CICs, circulating immature cells; G-CSF, granulocyte-colony stimulating factor; HGB, haemoglobin; PB, peripheral blood; WBC, white blood cell.
prepared 10-14 days before aphaeresis and transfused during BM collection.

CD34 + cell counts
PB cell counts before aphaeresis and BM harvest and PBSC and BM component counts were obtained using an automated cell counter (Model SE-9000; Sysmex Corporation, Kobe, Japan). White blood cells (WBCs), circulating immature cells (CICs), platelets and CD34 + cells were counted for analysis. CICs were defined as circulating myeloblasts, promyelocytes, myelocytes, metamyelocytes and erythroblasts detected by morphology in a PB smear; their counts were calculated by multiplying the percentage of the CICs with each donor's WBC count (Sierra et al., 1997). CD34 + content of collected products was quantitated by flow cytometric analysis as reported previously (Zhang et al., 2008). Briefly, 100 μL of each sample, containing (5-10) × 10 5 cells, was incubated for 10 min at room temperature with phycoerythrin [PE; (HPCA-2)]-conjugated anti-CD34 + monoclonal antibody (moAb), fluorescein isothiocyanate-conjugated anti-CD33 moAb and PerCP-conjugated anti-CD45 moAb (Becton Dickinson, Mountain View, CA, USA), whereas the control reagent contained a nucleic acid dye, PE-labelled IgG1 and PerCP-labelled anti-CD45 antibody. After red cell lysis with lysing solution, the CD34 + cell count was performed using a flow cytometer (FACScan, Becton Dickinson) and a ProCOUNT software package (BD) for each sample. At least 80 000 CD45 events and 1500 BD TruCOUNT tube beads were collected in each sample. Samples were analysed according to the gating strategy as described previously (Sierra et al., 1997). Nucleic acid dye vs. side-scattering cytogram (SSC) and CD45 PerCP vs. SSC dot plots were used to exclude cellular debris and any event with bright CD45 or high SSC, which are not characteristic of progenitor cells. The CD34 + cell population was defined in the nucleic acid dye vs. the CD34 + PE of cells gated from previous additive regions. The absolute number of CD34 + cells in the sample was determined by dividing the number of CD34 + cellular events by the number of fluorescent bead events and then multiplying this by the bead concentration. Stem cell yield was defined as the number of the CD34 + cells kg −1 of recipient BW. A final CD34 + cell count of <2 × 10 6 cells kg −1 (recipient BW) was considered a poor yield in our outcome analysis.

Statistical evaluation
Descriptive statistics, including mean and range, were calculated, and age, body mass index (BMI) and aphaeresis and BM products for male and female donors and recipients were compared using the two-sample t-test and the χ 2 test. BMI was defined as the individual's BW in kilograms divided by height squared in metres (kg m −2 ). The means of procedural variables, including actual flow rate, processed volume and CD34 + cell yield for PB and BM were compared using twosample t-tests based on all data and by stratified gender group separately. Multiple regression was used to analyse the correlations among potential factors, including donor characteristics, procedural variables and CD34 + cell yields. The risk for poor CD34 + cell yields was evaluated using a logistic regression model. A test was considered statistically significant when the P value was under 0·05. All analyses were performed using the spss 13·0 statistical package (SPSS Inc, Chicago, IL).

Factors affecting PBSC yield
Considering the influence of gender, male donors had significantly higher mean total CD34 + cell count (P < 0·0001) and cell yield (P < 0·0001) compared with female donors. The male donors had higher preaphaeresis haemoglobin (HGB) and donor cell dose but lower pre-aphaeresis PLT counts, pre-aphaeresis WBC count and age. Processed volume per kilogram of donor BW, actual flow rate of aphaereses, BMI, G-CSF dose, total processed volume and pre-aphaeresis CIC counts and recipient cell dose did not differ between the genders.
In multiple regression analysis, factors that significantly correlated with CD34 + cell yields were donor age, BMI, sex, flow rate and pre-aphaeresis WBC and CIC counts. Age and flow rate were negatively correlated with cell yield, whereas BMI and pre-aphaeresis WBC and CIC counts were positively correlated with cell yield. G-CSF dose and total processed blood volume did not affect the number of CD34 + cells kg −1 recipient BW. Regarding the OR of each factor for poor CD34 + cell yield (<2 × 10 6 kg −1 recipient BW), female gender and older age were significantly associated with increased risk based on logistic regression analysis. There was a trend of decreased risk for poor yield in donors with higher pre-aphaeresis laboratory data.

Factors influencing BM stem cells yield
Considering the influence of gender, male donors had significantly higher mean total CD34 + cell count (P < 0·0001) and cell yield (P < 0·0001) before and after management compared with female donors. The male donors had higher pre-BM collection HGB and donor and recipient cell doses before and after management but lower pre-BM collection PLT counts and age. Total processed volume, BMI, G-CSF dose, pre-BM collection WBC and CIC counts did not differ between the sexes.
In multiple regression analysis, factors that significantly correlated with CD34 + cell yield were donor age and pre-BM collection CIC counts. Age was negatively correlated with cell yield, whereas pre-BM collection CIC counts were positively correlated with cell yield. G-CSF dose, total processed BM volume and pre-BM collection WBC did not affect the number of CD34 + cells L −1 of processed BM.
Regarding the OR of each factor for poor CD34 + cell yield (<2 × 10 6 kg −1 recipient BW), older age was significantly associated with increased risk based on logistic regression analysis. There was a trend of decreased risk for poor yield in donors with higher pre-BM collection laboratory data.

Procedure tolerability
All donors tolerated G-CSF administration well. No donor required a reduced or increased G-CSF dosage because of a high leukocyte count on the third day of mobilisation. The most common symptoms during mobilisation were bone pain and myalgia. These were resolved with naproxen.
All donors tolerated the aphaeresis procedure. Symptoms related to hypocalcaemia, such as numbness and muscle cramps, were the most common adverse events; these resolved with calcium gluconate infusion and did not necessitate stopping the procedure. No major morbidity or mortality related to catheterisation was noted. There were no instances of bleeding complications, vascular access problems, or severe thrombocytopaenia during the aphaeresis sessions.
All donors tolerated the procedure of BM collection. There was no bleeding, low blood pressure or other symptoms. No other adverse diathesis was noted.

DISCUSSION
HLA-haploidentical/mismatched donors offer several advantages: (i) immediate donor availability for virtually all transplant candidates; (ii) ability to select the best of many relatives on the basis of age, infectious disease status and natural killer cell alloreactivity; (iii) controlled graft composition; (iv) immediate access to donor-derived cellular therapies if required after transplantation. Furthermore, for nearly all patients who face graft rejection, HLA-haploidentical/mismatched transplantation offers the advantage of either another family member who is immediately available as an alternative donor or even a second graft from the original donor (Aversa et al., 2008). However, haploidentical/mismatched HSCT with PBSCs or BM alone remains difficult with high GVHD and immune reconstitution (Passweg, 2006). To overcome these problems, the unmanipulated HLA-haploidentical/mismatched related HSCT was performed with combined G-PBSCs and G-BM because of higher volume of collected stem cells and GVHD and GVL regulations (Huang et al., 2004(Huang et al., , 2007Lu et al., 2006;Dong et al., 2007;Chen et al., 2009). Previous studies used different settings for aphaeresis equipment and different protocols, varied type and dosages of G-CSF and various parameters for outcome analysis and reported different final results from unrelated healthy donors (Anderlini et al., 1997;de la Rubia et al., 2002;Diaz et al., 2003;Ikeda et al., 2004;Kozuka et al., 2004;Lysak et al., 2005;Suzuya et al., 2005;Ings et al., 2006;Vasu et al., 2008). However, little information is available about the factors predicting the outcome of PBSC collection and BM harvest from unmanipulated HLA-haploidentical/mismatched related donors.
Final infused cell dose is frequently used as the outcome indicator when analysing the effect of donor characteristics on PBSC harvest. Today, most harvest centres perform at least two aphaereses if the first product does not achieve the target cell dose. The repeated aphaereses may mask the impact of donor factors on stem cell harvest with different numbers of aphaereses among donors and lead to bias due to the differences in criteria for a second aphaeresis among institutions. In addition, most previous studies have adjusted the CD34 + cell dose based on donor/recipient BW or on the volume of blood processed to standardise the yields among donors. Consequently, using the final number of CD34 + cells collected as the dependent variable may be inappropriate (Vasu et al., 2008). The number of CD34 + cells L −1 of processed blood volume collected was used as a factor to identify the stem cell yield and this should also be investigated (Sierra et al., 1997). In this study, we comprehensively analysed the factors that affect the stem cell yield with CD34 + cells kg −1 recipient BW and CD34 + cells L −1 of processed blood volume collected, which may compensate the deficiency only used alternative methods. Our results showed that the two methods have similar findings for the poor yield, which indicated that any method could be used to analyse the aphaeresis yield for unmanipulated HLA-haploidentical/mismatched related HSCT with combined G-PBSCs and G-BM.
There is a well-documented correlation between PB CD34 + count and CD34 + cell yield, although not always observed in healthy PBSC donors (Moncada et al., 2003). Other parameters that have been reported to predict CD34 + yield of aphaeresis are circulating WBC counts, BMI and the presence of immature myeloid forms (myelocytes, metamyelocytes, etc.) in the PB (Cassens et al., 2004;Kozuka et al., 2004;Suzuya et al., 2005). In our study, we found that the BMI, preaphaeresis WBC and CIC counts have significant impact on both CD34 + cells kg −1 recipient and CD34 + cells L −1 of processed blood volume collected.
Increasing donor age was associated with a modest negative effect on CD34 + mobilisation response (Anderlini et al., 1997). Similar report showed that older age is associated with poor mobilisation and concluded that larger, male donors younger than 55 years would be preferable (Ings et al., 2006). In contrast, other groups have reported that age is not a significant predictive factor (Miflin et al., 1996). In murine studies designed to determine whether aging is associated with changes in stem cell pool size or altered progenitor cell response to cytokines, the study found that aged mice exhibit better mobilisation responses to G-CSF and that their haematopoietic progenitor cells were characterised by reduced adhesion to marrow stroma (Xing et al., 2006). In our study, the results showed that age was negatively correlated with CD34 + yields.
Blood cell separators separate the different blood components on the basis of their individual densities by centrifugation. Stem cells are found within the buffy coat, which consists of the WBCs and platelets, as well as in the top of the RBC layer. The buffy coat is detected by the optic device. When the buffy coat is detected, valves are opened to transport the PBSCs into the storage bag (Moog, 2004). The relationship between the flow rate and the results of stem cell harvest is reported by few authors. They showed that a lower blood withdrawal rate was associated with a higher MNC yield and CD34 + cell yield (Suzuya et al., 2005;Wang et al., 2008). Our study also indicates that the yield of cells correlated negatively with flow rate. The cause for the results may be that entire time was utilized with low flow rate to separate the blood components and to detect the buffy coat by the optic device. However, the flow rate was not clearly defined or even described in previous studies and in our studies of factors affecting stem cell collection outcome. An optimal flow rate may need to be confirmed in a prospective setting.
The PB CD34 + cell count in healthy subjects under unstimulated conditions is very low, but it increases 15-to 35-fold following 4-5 days of G-CSF administration (Rhodes & Anderlini, 2008). The outcome of stem cell harvest depends on two steps: mobilisation and harvest. But few have described procedural factors affecting the outcome of PBSC collection. Some recent studies have focused on the effect of large-volume leukaphaeresis (LVL) (Wang et al., 2008). Although LVL may indeed increase final stem cell yield, other possible procedural factors, such as flow rate and circulation access, may also increase yield. In this study, we found that the dosage of twice daily G-CSF administration and the total processed blood volume did not affect the CD34 + yield.
PLT counts 1 day prior to or on the day of stem cell harvest were associated with CD34 + harvest (Suzuya et al., 2005;Tomblyn et al., 2005). The association of higher baseline PLT counts with improved CD34 + mobilisation may be related to common pathways of thrombopoiesis and progenitor cell mobility. Increased plasma levels of stromal-derived factor-1 have been shown to enhance human thrombopoiesis and mobilise human colony forming cells in non-obese diabetic/severe combined immunodeficiency mice (Perez et al., 2004). In addition, CD34 + cells may exhibit changes similar to platelet-derived microparticles during G-CSF mobilisation (Nomura et al., 2004). However, the CD34 + yields were not associated with the pre-aphaeresis PLT counts in this study.
The effect of gender on stem cell yield in our study disagreed with those of previous studies (Engelhardt et al., 1999;de la Rubia et al., 2002;Kozuka et al., 2004;Lysak et al., 2005). Two recent large studies found higher total final CD34 + cell count and cell yield in male donors (Fischer et al., 2005;Ings et al., 2006). The significantly lower numbers of post-G-CSF circulating CD34 + cells in female donors were detected by another study (Vasu et al., 2008). We noted that female gender had a lower mean harvest outcome and a higher risk for poor yield in our cohort. The actual underlying reasons for the relatively poor mobilisation by G-CSF in female donors remain to be elucidated. The development of newer mobilising agents, such as CXCR4 antagonists, should be explored with regard to the CD34 + cell mobilisation in female donors.
The final adequacy of the product is determined by the CD34 + cell count of the PBSC aphaeresis product. While the collection target is usually 4-5 × 10 6 kg −1 recipient BW, the minimum number of CD34 + cells required to ensure prompt engraftment after allogeneic transplantation is probably in the range of 2 × 10 6 kg −1 recipient BW (Rhodes & Anderlini, 2008). A small number of donors (up to 5%) prove to be poor mobilisers and fail to reach even the minimum CD34 + cell dose despite two or more collections (Anderlini et al., 1999). Previous studies have shown delayed engraftment at CD34 + doses less than 8 × 10 6 kg −1 BW (Koh & Chao, 2008). Most physicians would usually target for 'megadose' of stem cell (>10 × 10 6 CD34 + cells kg −1 recipient BW) from the donor while planning for HLA-haploidentical transplants, and this can place considerable demand on both the donors and the phaeresis service for the following reasons: (i) The high graft contents are easily achieved in children but can be a major obstacle in adults. (ii) The immobility from long hours of phaeresis can be exhausting for the donors and the procedure is often associated with significant aphaeresisrelated adverse effects. (iii) For the aphaeresis and stem cell processing laboratory staff, the procedures involved can be time-consuming and labour-intensive (Lang et al., 2004). The BM harvest was performed in our study and was found that age was negatively correlated with CD34 + yield, whereas pre-BM collection CIC counts were positively correlated with CD34 + yield, and G-CSF dose, total processed BM volume and pre-BM collection WBC did not affect the CD34 + yield.
Most common side effects are mild-to-moderate, with bone pain, myalgia, malaise and nausea being the most frequently occurring during mobilisation (Stroncek et al., 1996). These side effects are readily treated with common analgesics, such as acetaminophen. G-CSF also causes a mild-to-moderate decrease in PLT count from baseline in approximately 25% of healthy allogeneic donors (Tassi et al., 2005). The aphaeresis procedure itself causes an additional and more substantial PLT depletion and transient G-CSF-induced hypercoagulability in PBSC donors (Falanga et al., 1999;Topcuoglu et al., 2004). On the other hand, the coagulation system may change because of the use of large anti-coagulation drugs. Hence bleeding appears to be common; however, no bleeding was observed in our study. In this study, 1-2 units of donor's own blood were prepared and infused back into the donor during BM harvest because a significant amount of blood was taken with the marrow, which prevented low blood pressure during harvest and minimised the need for a blood transfusion from an outside source after the harvest and stopped the side effect of allotransfusion.
In conclusion, we found that female gender, older age and higher flow rate were significantly associated with increased risk for poor harvest of PB. A higher BMI may decrease the risk. Age was negatively correlated with CD34 + cell yield of BM, whereas pre-BM collection CIC counts were positively correlated with CD34 + cell yield. Therefore in a donor with multiple high-risk characteristics (e.g. an older female with a lower BMI) for poor CD34 + cell yield, strategies for increasing CD34 + cell yield must be considered besides repeated aphaeresis. BM stem cells could be an alternative source in these donors because our data showed no correlation between gender and CICs of PB in BM harvest.