Fully automated, clinical-grade bone marrow processing: a single-centre experience

Abstract

Background

Clinical grade processing of harvested bone marrow is required in various clinical situations, particularly in the management of ABO mismatching in allogeneic haematopoietic stem cell transplantation (HSCT) and in regenerative medicine.

Material and methods

We report a single-centre experience using a fully automated, clinical grade, closed system (Sepax, Biosafe, Switzerland). From 2003 to 2015, 125 procedures were performed in our laboratory, including buffy-coat production for HSCT (n=58), regenerative medicine in an orthopaedic setting (n=54) and density-gradient separation in a trial for treatment of critical limb ischaemia (n=13).

Results

Buffy coat separation resulted in a median volume reduction of 85% (range, 75–87%), providing satisfactory red blood cell depletion (69%, range 30–88%) and a median recovery of CD34 cells of 96% (range, 81–134%) in the setting of allogeneic HSCT. Significantly greater volume reduction (90%; range, 90–92%) and red blood cell depletion (88%; range, 80–93%) were achieved by the new SmartRedux software released for Sepax2, validated in the last eight allogeneic HSCT. The density gradient separation programme resulted in complete red blood cell depletion associated with high CD34 recovery (69%; range, 36–124%). No reactions related to the quality of the product were reported. Time to engraftment following allogeneic HSCT was in the normal range. No cases of microbiological contamination related to the manipulation were reported.

Discussion

Clinical grade, automated bone marrow manipulation with Sepax was shown to be effective, giving operator-independent results and could be used for a broad range of clinical applications.

Introduction

Haematopoietic stem cell transplantation (HSCT) is a medical procedure commonly used for the treatment of a wide variety of haematological malignancies, metabolic disorders and immunodeficiency states¹.

Different sources of stem cells can be used: bone marrow (BM), mobilised peripheral blood stem cells and umbilical cord blood. Peripheral blood stem cells have replaced BM in the autologous setting, while about 30% of allogeneic transplants still use BM as the source of stem cells, probably because of the higher incidence of chronic Graft-versus-Host disease following peripheral blood stem cell transplantation^2,3. In the setting of allogeneic HSCT, 30% of transplants from HLA-identical siblings and 50% from unrelated donors are red blood cell (RBC) incompatible^4,5.

The immuno-haematological consequences of major and minor ABO incompatibilities have been summarised by Rowley et al. and include acute and delayed haemolysis, delayed RBC recovery and pure red cell aplasia⁶.

The risk of reactions, which can have an abrupt onset and may be fatal, is lowered by graft processing and proper blood component support. Standard procedures for ABO incompatible transplants consist of RBC and/or plasma depletion⁷ carried out by apheresis devices, density-gradient separation and simple centrifugation, possibly associated with the use of sedimentation agents, such as hydroxyethyl starch^8,9. Centrifugation of the graft is performed in order to enable collection of the total nucleated cell layer at the interphase between the plasma and RBC pellet (buffy coat), or the overall cellular pellet after plasma removal (plasma depletion). An algorithm for the management of RBC incompatibility was proposed⁶ with the aim of standardising the graft manipulation.

Processing of intermediate/small volumes of BM harvests is also requested for graft preparation in regenerative medicine protocols. Autologous BM mononuclear cells (MNC) have been safely used in different clinical experimental trials for the treatment of critical limb ischaemia¹⁰ and myocardial infarction¹¹ and buffy coat has been used in orthopaedic conditions¹². According to the clinical target, the volume of the harvested BM and the final characteristics of the product may vary largely and flexible processing methodologies are, therefore, required.

Manipulation of BM is a key factor in the transplantation process and may influence both engraftment and overall survival^7,9. Standardisation of cell processing would decrease both variability of the final product quality and the need for specialised staff training. Here we report a single-centre experience of the use of a fully automated, clinical-grade, closed system (Sepax, Biosafe, Eysins, Switzerland) for BM processing in different clinical settings.

Materials and methods

Cell processing: the Sepax system

BM was processed in either a Sepax S-100 or Sepax 2 device (Biosafe), managed through software specifically developed for different clinical targets, in closed, disposable kits. The Sepax cell processing system uses a rotating syringe technology that allows separation of blood components through rotation of the syringe chamber. Blood components (plasma, buffy coat and red cells) are detected by an optical sensor and transferred into different output bags by diverting the output flow of the syringe piston.

Operating software

Non-density gradient separations in ABO-incompatible transplant procedures and in the orthopaedic setting were performed with either generic volume reduction (GVR) or SmartRedux software (Biosafe). The GVR protocol enables the collection of BM buffy coat or plasma-free BM. CS-490 disposable kits were used for all the procedures. As the volume of the syringe chamber is 220 mL, multiple cycles of centrifugation must be performed for larger BM volumes. The GVR protocol allows initial product volumes from 50 mL to 880 mL, so multiple procedures are required for larger BM volumes. The protocol is user-adaptable and various parameters can be adjusted in order to maximise cell recovery or decrease RBC contamination of the final product. The processing time for 880 mL of BM is about 60 minutes. A further evolution of GVR software is SmartRedux, available only with the Sepax2 device: this software can deal with an input volume ranging between 30 and 3,300 mL and a large volume product can, therefore, be processed in a single procedure with one kit. The time for SmartRedux processing of 1 L of BM ranges between 20 and 210 minutes, according to the final product (plasma depletion or buffy coat). Once the operator has set the desired parameters, the whole procedure is carried out in a fully automated manner. Either saline or autologous plasma can be added to the product in order to decrease the product density before infusion into the patient.

The density-gradient based separation software (NeatCell, Biosafe) is designed to manage automated separation of BM on a density gradient of 1.077 g/mL, in order to obtain a purified MNC fraction. It was used either in regenerative medicine protocols, when a purified MNC fraction was required, or in the case of major incompatibility with a high titre of natural anti-ABO antibodies. The separation process is carried out in a specific closed, disposable kit (CS-900). After the separation, the product is washed two or three times, according to the operator’s settings, in saline with added human serum albumin. The input volume ranges between 30 and 120 mL while the final output volume can be set up by the operator in a range between 8 and 45 mL. The processing time with three washing cycles is about 90 minutes.

Patients

Haematopoietic stem cell transplantation

Between 2003 and 2013, 24 patients with haematological malignancies received a BM transplant in our facility in Florence (Italy): 19 had allogeneic ABO mismatching (11 major, 4 minor and 4 double) and five required volume reduction of autologous BM before cryopreservation of the product following failure of peripheral blood stem cell mobilisation. Between December 2014 and January 2016, eight more patients with haematological malignancies underwent allogeneic transplantation with ABO mismatching (5 major and 3 minor). The patients’ characteristics are reported in Table I. Engraftment was assessed as time (days) to reach a neutrophil count of 0.5×10⁹/L.

Table I.

Characteristics of the allogeneic and autologous HSCT recipients.

Patients’ characteristics
Allogeneic HSCT		Autologous HSCT
Diagnoses	n.	Diagnoses	n.
Acute myeloid leukaemia	14	Acute myeloid leukaemia	3
Acute lymphoblastic leukaemia	9	Multiple myeloma	1
Non-Hodgkin’s lymphoma	2	Non-Hodgkin’s lymphoma	1
Myeloproliferative neoplasm	1
Myelodysplastic syndrome	1
Type of donor
Matched sibling donor	4
Matched family donor	7
Matched unrelated donor	16
ABO-mismatch (donor/recipient)
A+/B+	1
A+/O+	7
A+/O	3
A /O	1
B+/O+	4
B+/A+	3
AB+/O+	1
O+/AB+	1
O+/A+	2
O /A	1
O+/A	1
O+/B+	2

Patients are grouped based on diagnosis, type of donor and ABO mismatch. HSCT: haematopoietic stem cell transplantation.

Regenerative medicine

In the same period, autologous BM processing was carried out within approved trials of regenerative medicine: volume reduction was achieved by either simple buffy-coat separation (GVR programme) in a protocol of bone regeneration in the orthopaedic setting (n=54) or density-gradient mononuclear cell separation (NeatCell) in a trial on the treatment of critical limb ischaemia (n=13).

Bone marrow harvesting

BM units were collected in our centre or in other accredited centres for mismatched unrelated donor marrow collection, according to the standard procedure of each centre. BM was harvested by multiple aspirations from the iliac crest and cells were collected into transfer bags containing ACD and heparin. The volume collected ranged between 150 and 1,727 mL, depending on the clinical application.

Assessment of bone marrow content

White blood cell (WBC) and RBC counts and haematocrit were determined before and after processing using an automated haematology analyser (XS-1000i, Sysmex Corp., Kobe, Japan). CD34+ cell count, MNC quantification and viability as well as WBC viability were determined by immunofluorescence analysis (FACScanto, BD Pharmingen, San Diego, CA, USA) using the ISHAGE protocol with the single platform technique^13–15. 7-aminoactinomycin D (BD Pharmingen) was used to exclude dead cells.

Haematopoietic progenitor cell assays were performed in duplicate, with the methylcellulose-based medium containing recombinant cytokines (MethoCult GF H4434; Stem Cell Technologies, Vancouver, Canada) as previously described¹⁶. Total colony-forming cells, defined as the total number of colonies, regardless of their lineage, were used to calculate clonogenic potential.

The number of fibroblast colony-forming units (CFU-F) was determined as previously described¹⁶, and was used as a surrogate marker for mesenchymal stem cells progenitors. Before and after separation, 1×10⁶ total nucleated cells were plated in duplicate in 100 mm Ø Petri dishes. After 14 days, the dishes were fixed with methanol and stained with Giemsa; visible colonies formed by 50 or more cells were counted and reported as the number of CFU-F/10⁶ seeded total nucleated cells.

Statistical analysis

Data were extracted retrospectively from our leucapheresis/BM processing database. Data are expressed as mean ±SD or median (range) and comparisons were carried out using the Student’s t-test or the non-parametric Mann-Whitney test.

Results

Volume reduction

Initial and final BM volumes as well as percentage volume reduction (median and range) are reported in Table II. Data are divided for allogeneic HSCT, autologous HSCT, buffy coat and density-gradient separation for regenerative medicine. In all cases the BM volume reduction was greater than 72% (Table II).

Table II.

BM volumes before and after processing and percentage volume reduction for each protocol.

Patients	N	Input volume (mL)	Output volume (mL)	Volume reduction (%)
Allogeneic HSCT	19	1,221 (773–1,727)	200 (120–300)	85 (75–87)
Autologous HSCT	5	1,047 (1,021–1,129)	150 (119–153)	86 (85–88)
BC/RM	54	210 (150–300)	30 (22–45)	86 (79–90)
DGS/RM	13	300 (200–360)	42 (27–51)	86 (73–90)

Data are expressed as median (range). BM: bone marrow; HSCT: haematopoietic stem cell transplantation; BC: buffy coat; DGS: density-gradient separation; RM: regenerative medicine.

Bone marrow processing for ABO mismatch

When the GVR protocol for RBC and plasma removal (buffy coat extraction) was used, multiple procedures were needed if the initial volume exceeded 880 mL; overall 50 procedures were carried out. For allogeneic HSCT the median CD34+ cell recovery was 95.9% (range, 81.2–134.3%); CD34+ recovery was not available for autologous HSCT. The median percentage WBC recovery was 86.0% (range, 76.0–104.2%) in allogeneic HSCT and 74.0% (range, 57.1–76.3%) in autologous HSCT (Figure 1). Reductions in packed RBC volume with the GVR protocol in the HSCT setting are reported in Table III. The median reduction in RBC volume was 69% (range, 30–88%) for allogeneic transplantation and 73% (range, 59–85%) for autologous transplantation.

Figure 1.

Percentage recovery of CD34 cells and WBC for the GVR protocols for BM buffy coat extraction for infusion in allogeneic HSCT, autologous HSCT and in regenerative medicine for patients with orthopaedic disorders (BC/RM).

Each bar represents the median, quartiles, minimum and maximum values. WBC: white blood cells; GVR: generic volume reduction; BM: bone marrow; HSCT: haematopoietic stem cell transplantation; BC: buffy coat; RM: regenerative medicine.

Table III.

RBC volumes before and after processing and percentage RBC reduction for HSCT and density-gradient separation for regenerative medicine protocols.

	RBC volume before processing (mL)	RBC volume after processing (mL)	RBC volume reduction (%)
Allogeneic HSCT	349 (193–525)	116 (36–217)	69 (30–88)
Autologous HSCT	344 (225–355)	93 (49–99)	73 (59–85)
DGS/RM	67 (32–107)	0.4 (0–0.6)	100 (99–100)

Data are expressed as median (range). RBC: red blood cell; HSCT: haematopoietic stem cell transplantation; DGS: density-gradient separation; RM: regenerative medicine.

The SmartRedux protocol was used with the Sepax 2 machine, allowing an input volume up to 3,300 mL and enabling the final volume to be set. This software was validated in five allogeneic HSCT with major and three with minor ABO mismatching; the data are summarised in Table IV, and compared with data obtained when the GVR programme was used. The SmartRedux protocol was associated with greater volume reduction (90%; range, 90–92% vs 85%; range, 75–87%: p<0.05) and RBC volume reduction (88%; range, 80–93% vs 69% range, 30–88%, p<0.05), associated with full CD34+ recovery, resulting in a median final packed RBC volume less than 50 mL.

Table IV.

BM processing for allogeneic HSCT: GVR vs Smartredux protocol.

	GVR (n=19)	SmartRedux (n=8)	p-value
Pre-processing	1,221	1,128
volume (mL)	(773–1,727)	(87–1,440)
Post-processing	200	110
volume (mL)	(120–300)	(75–150)
Volume reduction	85	90
(%)	(75–87)	(90–92)	p<0.05
CD34 recovery %	96	105
	(81–134)	(82–114)
WBC recovery %	86	92
	(76–104)	(87–111)
Pre-processing RBC	349	380
volume (mL)	(193–525)	(234–589)
Post-processing RBC	116	44
volume (mL)	(36–217)	(24–50)
RBC volume	69	88
reduction (%)	(30–88)	(80–93)	p<0.05

BM volumes before and after processing and percentage volume reduction, percentage of CD34 and WBC recovery, RBC volumes before and after processing and percentage RBC volume reduction are reported for each protocol. Data are expressed as median and range. The final volume in the GVR protocol is the total output of the single procedures used to process grafts exceeding 880 mL. SmartRedux software always allowed single-step procedures.

Statistical comparisons were carried out with the Mann-Whitney U test. BM: bone marrow; HSCT: haematopoietic stem cell transplantation; GVR: generic volume reduction; WBC: white blood cells; RBC: red blood cells.

Bone marrow buffy coat extraction for regenerative medicine

The GVR programme was applied for a bone regeneration protocol in the orthopaedic setting in 54 patients. The median volume of autologous BM processed was 210 mL (range, 150–300 mL). The median CD34+ and WBC recovery was 79.4% (range, 41.1–109.2%) and 50.8% (range, 23.0–84.4%), respectively (Figure 1). The final, median reduced volume was 30 mL (range, 22–45 mL) (Table II).

Mononuclear cell separation for regenerative medicine

The density-gradient separation procedure (NeatCell protocol) resulted in a satisfactory recovery of CD34+ cells (69%; range 36–124%), MNC (67%; range, 17–119%) and colony-forming cells (49%; range, 9–124%), but a quite low CFU-F recovery (30%; range, 8–50%) according to our previous data¹⁶. The procedure resulted in a median polymorphonuclear cell reduction of 99.8% (range, 99.7–100%); complete RBC depletion (100%) was also achieved (Table III), therefore making this procedure suitable for BM processing in the case of allogeneic HSCT with major ABO mismatching and a high titre of natural antibodies or irregular anti-ABO antibodies.

Clinical outcome

Neutrophil engraftment was achieved at a median of 23 days (range, 13–31 days) for allogeneic HSCT and 19 days (range, 9–31 days) for autologous HSCT. One patient did not engraft after a haploidentical HSCT as rescue for failed engraftment of a cord blood transplant; the patient died of sepsis 17 days after the rescue transplant.

No infusion-related side effects were reported; in particular, no transfusion reactions occurred in the first group (ABO-mismatched allogeneic HSCT) except in one patient who developed a reaction after the infusion of 20 mL of the product, possibly due to the presence of irregular antibodies. The graft was then rescued by MNC separation with the NeatCell protocol and the infusion was carried out without any side effects. No microbiological contamination due to the BM manipulation process was reported.

Discussion

BM transplantation is commonly performed across ABO mismatching between donor and recipient. Both early (during graft infusion) and delayed (during engraftment) haemolytic transfusion reactions are expected complications. The presence of host anti-donor isoagglutinins (major incompatibility) has an important clinical impact as it can delay RBC recovery and increase transfusion requirements. Although ABO mismatching does not affect neutrophil and platelet engraftment, conflicting data have been reported regarding its impact on Graft-versus-Host disease and overall survival⁶.

The volume of infused incompatible RBC predicts the likelihood of symptomatic reactions. Nevertheless, there is no “safe” quantity of red cells below which haemolytic transfusion reactions will not occur as allograft recipients can experience serious haemolytic reactions to even a small volume of RBC, typically less than 10 mL^17,18.

In the majority of transplant centres BM is processed before transplantation in the case of ABO major incompatibility with the goal of removing the donor’s RBC and thereby infusing the graft safely. However, some centres remove erythrocytes only in the case of high recipient titres (>1:256). Attempts to standardise the management procedures have been made by different institutions^8,19,20.

Currently several strategies are adopted for reducing the RBC content of BM harvests: (i) gravity sedimentation after adding hydroxyethyl starch or dextran to the BM to agglutinate the RBC, resulting in faster sedimentation of these cells than the WBC^21–23; (ii) buffy coat separation using cell separators and apheresis devices^9,24–26; and (iii) separation on a density gradient⁸. A combination of two methods is also used, with the addition of hydroxyethyl starch to BM in a blood cell separator⁸. The goal is to ensure high nucleated and CD34+ cell recovery rates as well as a good depletion of RBC. Different methodologies and devices for BM processing have been compared recently^27,28.

We report here the experience of our centre in 24 patients who underwent ABO-mismatched allogeneic HSCT for haematological malignancies. We found that the Sepax platform enables plasma removal and RBC depletion, while ensuring high recovery of WBC and CD34+ cells by a non-density gradient process. Such methodology is routinely used in our centre for BM processing prior to ABO-mismatched allogeneic transplantation and resulted in a satisfactory recovery of WBC (median 86%; range, 76–104%) and CD34+ cells (median 96%; range, 81–134%). Our device achieved greater WBC recovery and a CD34+ cell recovery comparable to other devices currently used for BM manipulation for ABO-mismatched transplantation, such as the Fresenius COMTEC (WBC: 37%; range, 23–53%; CD34+ cells: 96.2%; range, 80–100%)²⁹, Cobe SPECTRA (WBC: 33.7±12.2%; range, 10.3–76.3%; CD34+ cells: 82.2±21.1%; range, 26.7–159.8%)⁹ and AMICUS (WBC: 44±16%; CD34+ cells: 70±18%)²⁴.

The clinical efficacy of the grafts was in line with that of historical controls and no microbiological contamination was reported in the final infused product.

The analysis of case series reported in the literature suggests that the volume of packed RBC in the case of a major ABO-mismatched transplant should be less than 50 mL¹⁷; however, data are very patchy and no safe volume could be extrapolated. While the majority of our patients had an ABO titre >1:8, in our experience a volume of packed RBC around 100 mL was not associated with any infusion-related side effects and the haematopoietic recovery was similar to that observed in patients transplanted with profoundly RBC-depleted grafts⁹. A transfusion reaction occurred in one patient after the infusion of 20 mL of graft, containing a few mL of packed RBC; the reaction was possibly related to circulating irregular anti-RBC antibodies, and the graft was infused without any side effects after further manipulation with the NeatCell programme. This density gradient-based BM processing system is able to recover progenitor cells (CD34+ recovery 69%; range, 36–124%) and eliminate RBC completely, with a performance comparable to that of other separation devices and protocols²⁸.

According to the policy of our BM transplant centre, plasma exchange was always performed before a conditioning regimen if the anti-ABO antibody titres were higher than 1:256 before starting the conditioning regimen. An infusion protocol, aimed at limiting the speed of graft infusion in the case of ABO mismatching is also included in our operating procedures. These two last strategies might have played a role in the lack of infusion-related adverse events in our series of patients. The final packed RBC volume with GVR software was higher than that reported with other methodologies^27,28, whilst the new SmartRedux software maintained the same performance in terms of WBC and CD34+ cell recovery, achieving a median volume RBC reduction of 88% (range, 80–93%). The final packed RBC volume was reduced to ≤50 mL, which has been reported to result in a lower occurrence of haemolytic reactions^6,17. However preliminary data obtained with the latter protocol need to be confirmed in a larger series of patients.

The Sepax system was also used in our centre for BM manipulation within approved experimental clinical protocols of regenerative medicine. Purified BM MNC were used for infusion in the gastrocnemius muscle in patients with critical limb ischaemia. The NeatCell process resulted in high MNC and CD34+ recovery rates (about 70%) as well as complete depletion of RBC and polymorphonuclear cell reduction greater than 99%. No adverse reactions during intramuscular cell injection were reported for the 13 patients enrolled in the study. As previously mentioned, our results are in line with those of other studies using the NeatCell programme or other devices using Ficoll-based protocols such as COBE2991 and CliniMACS Prodigy²⁸.

Autologous BM buffy coat extraction was performed on small volumes of BM (range, 150–300 mL) for infusion in 54 patients with orthopaedic disorders. The cells were re-suspended with bone chips and autologous platelet gel before implantation. The procedure allowed stem cell concentration in a fixed volume (30 mL) in an intra-operative time frame; the processing time for such a volume is about 20 minutes.

Buffy coat separation in small volumes, such as those used for regenerative medicine, resulted in a lower WBC recovery (p<0.05) than that with large volume products. However, the recovery of both CD34+ cells and CFU-F was similar to that with large volume products and the difference was, therefore, related only to greater neutrophil loss, as previously reported¹⁶.

Our data support the safety and efficacy of the Sepax platform as a time-saving system to provide clinical-grade BM stem cells, as already demonstrated in other clinical settings¹⁰. In particular, the system is frequently used in the field of cord blood Banking and in stem cell laboratories without apheresis devices; it was recently shown to be effective in clinical-grade washing of thawed peripheral blood stem cells before autologous HSCT³⁰. Compared to other devices routinely used in apheresis collection units, the Sepax system is able to process smaller BM volumes (minimum 30 mL) and is easily manageable, therefore being suitable for direct use in the operating theatre²⁷.

Conclusions

In conclusion, standardisation of BM manipulation procedures is a key step in the transplantation process, facilitating the recovery of high quality products that determine the clinical outcome. BM manipulation with the Sepax system was shown to be effective, operator-independent and safe; all the validated protocols resulted in a high rate of haematopoietic stem cell recovery and could be used for a broad range of clinical applications with different stem cell sources.