A pre-clinical model of double versus single unit unrelated cord blood transplantation

Abstract

Cord blood transplantation (CBT) with units containing total nucleated cell (TNC) dose >2.5×10⁷/kg is associated with improved engraftment and decreased transplant-related mortality. For many adults no single cord blood units are available that meet the cell dose requirements. We developed a dog model of CBT to evaluate approaches to overcome the problem of low cell dose cord blood units. This study primarily compared double- versus single-unit CBT. Unrelated dogs were bred and cord blood units were harvested. We identified unrelated recipients that were dog leukocyte antigen (DLA)-88 (class I) and DLA-DRB1 (class II) allele-matched with cryopreserved units. Each unit contained ≤ 1.7×10⁷ TNC/kg. Recipients were given 9.2 Gy total body irradiation and DLA-matched unrelated cord blood with post-grafting cyclosporine and mycophenolate mofetil. After double-unit CBT, 5 dogs engrafted and 4 survived long term with one dominant engrafting unit and prompt immune reconstitution. In contrast, 0 of 5 dogs given single-unit CBT survived beyond 105 days (p=0.03, log-rank test); neutrophil and platelet recovery was delayed (both p=0.005) and recipients developed fatal infections. This new large animal model showed that outcomes were improved after double-unit compared to single-unit CBT. After double-unit CBT, the non-engrafted unit facilitates engraftment of the dominant unit.

Introduction

Cord blood transplantation (CBT) has emerged as an effective treatment for patients with malignant and non-malignant hematologic diseases. Clinical experience has shown that the infused cell dose and degree of human leukocyte antigen (HLA) matching of cord blood units are associated with engraftment and outcome. Several studies have shown that total nucleated cell (TNC) count and CD34+ cell doses above minimum thresholds are associated with improved engraftment and decreased transplant related mortality [1–4]. As a result, several transplant center guidelines recommended that the minimum threshold cell dose for a single cord blood unit is at least ≥ 2.5 – 3.0×10⁷ TNC/kg for a 5/6 or 6/6 HLA-matched unit (with low resolution typing at HLA-A and -B and high resolution at -DRB1) [5–7]. With greater HLA disparity, the threshold cell dose is even greater: the minimal cell dose for a single 4/6 HLA-matched unit is at least ≥ 5.0 × 10⁷ TNC/kg [6]. For many adults and older children no single units are available that meet these cell dose requirements. To overcome this limitation, investigators at the University of Minnesota introduced double unit cord transplantation in which two cord units are infused simultaneously to increase the total infused cell dose [8,9]. Nonrandomized clinical studies indicate that compared to low cell dose single unit CBT, double unit CBT appears to have increased engraftment and decreased transplant related mortality [8,10,11]. Many transplant centers now routinely use double unit CBT for adults and there is currently a randomized clinical trial in progress comparing single vs. double unit CBT in children. After double unit CBT, in the vast majority of cases, a single cord blood unit emerges as the sole source of long-term hematopoiesis. Although total nucleated cell (TNC), CD34+, CD3+ cell count, sex-mismatch, ABO blood group, HLA mismatch, and order of infusion have been evaluated, to date, no factors have been identified that consistently predict which unit will emerge as the dominant engrafting unit [12]. Recent data however suggest the unit with higher CD34+ cell viability at the time of thawing predicts the subsequent dominant engrafting unit, particularly when the non-engrafting unit has <75% CD34+ viability [13].

Because of limitations with the clinical trials of CBT, questions still remain as to the benefit of double unit CBT and the biologic determinants of engraftment following double unit CBT. To address these questions we aimed to develop a large animal model of CBT. Because of the long track record of successful translation of experimental findings in the outbred dog model to the clinic [14,15], we investigated if the dog model of CBT could begin to address questions of cell dose and major histocompatibility complex (MHC) barriers to engraftment and survival after high dose total body irradiation (TBI).

Material and Methods

Animals and Surgical Procedures

All procedures and research protocols were approved by the Institutional Animal Care and Use Committee of the Fred Hutchinson Cancer Research Center (FHCRC). Research and animal housing was conducted according to the principles outlined in the Guide for Laboratory Animal Facilities and Care prepared by the National Academy of Sciences and National Research Council. Dogs were raised at the FHCRC or obtained from commercial kennels licensed by the US Department of Agriculture. Adult female, outbred beagle and mixed breed mini-mongrel and hound dogs from which cord blood units were harvested weighed 8.0–38.0 (median, 10.4) kg and were 18 –122 (median, 23.5) months old. CBT recipient beagle and mixed breed dogs weighed 7.0 – 32.0 (median, 9.8) kg and were 9.5–22.0 (median, 14.4) months old. All dogs were examined at least twice daily. Recipient dogs were euthanized after transplantation when established clinical criteria were met for poor condition with infectious complications.

Details of the canine cord blood unit collection and characterization are described in the supplemental materials/methods section (available online). Briefly, at day 53 to 58 of gestation (estimated at 3 to 6 days before full term), gravid dogs were placed under general anesthesia and underwent a cesarean section procedure. There were 2 to 12 fetuses per gravid female; each fetus had a placenta that was separate and distinct from the other littermates. Throughout the fetal/cord blood collection procedure, each fetus was closely monitored for complete anesthesia. Due to the small size of the canine umbilical cord, blood collection from the isolated, clamped umbilical cord resulted in insufficient cell dose yield needed for subsequent transplantation experiments. The CD34 cell and colony forming unit (CFU) content of fetal jugular vein and umbilical cord blood were equivalent (Figure S.1, Tables S.1 and S.2). To minimize the number of animals used as cord blood donors, and to increase the cell dose yield from each fetus, we combined fetal and umbilical cord blood as the source of blood cells for all subsequent experiments. For this manuscript, CBT in dogs is defined as the transplantation of the combined fetal and umbilical cord blood.

Immediately after collection of each venous fetal/cord blood unit, aliquots of cord blood were collected for DLA typing and hematologic characterization. Immunophenotyping of blood was completed (Tables S.3 and S.4) [16]. After the cord blood unit cell dose was determined, the cells were cryopreserved.

DLA-Typing

Dog leukocyte antigen (DLA)-identical sibling units for CBT were chosen on the basis of complete family studies showing identity for highly polymorphic MHC associated class I and II microsatellite markers and by direct sequencing of DLA-DRB1 [17,18]. Initial experiments with DLA-identical siblings used recipient dogs from a prior litter of the same breeding pair. DLA-matched unrelated units for CBT were chosen on the basis of direct sequencing for both DLA DRB1 (class II) and DLA-88 (class I) allelic identity [18,19]. Breeding to generate the unrelated cord blood units was performed after the DLA type of the female and male was determined and dogs heterozygous for the most common DRB1 alleles, 00101, 00102, 00201, 00601, 00801, and 01501 were selected. The breeding pairs were unrelated by at least 6 generations. The CBT recipients were selected after DLA allele sequencing of adult dogs. Subsequently, DLA-matched unrelated recipient dogs were identified from randomly selected dogs that were unrelated to the cord blood donors by at least 6 generations in a separate, genetically diverse outbred colony.

Transplantation

Total body irradiation (TBI) 9.2 Gray (Gy) was delivered as a single fraction at 7 cGy/min from a 4 MEV (for the initial experiments with DLA-identical siblings units) and, more recently, with a 6 MEV linear accelerator (CLINAC 4/80 and CLINAC 600 C/D, respectively, Varian Associates, Palo Alto, CA). This is considered to be a myeloablative TBI dose [20,21]. CBT consisted of intravenous infusions of unwashed, thawed cord blood units each unit containing 0.3 to 1.7 ×10⁷ nucleated white blood cells per kilogram recipient body weight (TNC dose was obtained prior to cryopreservation and excluded nucleated red blood cells). The day of CBT was designated day 0. Cord blood unit CD34+ cells were analyzed for viability after thawing. Flow cytometric evaluation was performed using the International Society of Hematotherapy and Graft Engineering (ISHAGE) gating strategy [22] with minor modifications to maximize the detection of live and dead cells (Figure S.2) [13].

Recipients were given postgrafting immunosuppression for prevention of graft rejection and graft-versus-host disease (GVHD). Cyclosporine (CSP) 15 mg/kg orally (po) twice daily (BID) was given starting from day −3 to day +49. Mycophenolate mofetil (MMF) 5 mg/kg subcutaneous BID was given from day 0 starting after TBI to day +3, the dose was increased to 7.5 mg/kg BID on days +4 to 42. CSP dose was adjusted based on weekly pre-dose levels with a target range of 250–500 ng/mL. To prevent graft rejection among recipients of single unit unrelated CBT, the CSP dose was extended from days 50 to 98 at 7.5 mg/kg po BID and MMF was continued from days 43 to 84 at 3.75 mg/kg subcutaneous BID. Recombinant canine granulocyte colony stimulating factor (G-CSF) 5 µg/kg subcutaneous BID was initiated after TBI and continued until sustained recovery of absolute neutrophil count (ANC) >1000/µL for 3 consecutive days. Supportive care included prophylaxis with the oral antibiotic enrofloxacin from the day of TBI. Broader antibiotic coverage was administered with ceftazidime and vancomycin plus additional antibiotics as clinically indicated when ANC declined to below 500/µL and fever developed. Lactated Ringer’s solution 8–10 mL/kg intravenous was administered at least twice daily after TBI until there was complete clinical recovery (additional details are described in the Supplemental Data# 1: Materials and Methods section). Irradiated blood transfusions [23] were given either when platelet counts declined below 6 ×10⁹/L or when petechiae and ecchymoses of skin and mucous membranes were observed or when the hematocrit was <24%.

Assessment of engraftment and chimerism

The contributions of recipient and donor cells to peripheral blood and bone marrow were quantified by fluorescent variable number tandem repeat (VNTR) analysis [24]. Since some recipients received cells from 2 or 3 donor units, multiple loci were analyzed to fully distinguish the recipient and each donor unit for each transplant.

Immune reconstitution

In vitro immune functions of recipients were measured by mixed lymphocyte reaction (MLR), with a panel of three stimulators and autologous PBMC to generate a stimulation index [25]. In addition, peripheral blood CD3, CD4 and CD8 and CD21 phenotypes were determined by fluorescence cytometric analysis with fluorescein-conjugated CA17.6F9, and phycoerythrin conjugated CA13.1E4, CA9.JD3, and CA2.1D6, respectively [24]. Analysis was performed on a FACSCalibur or FACScan flow cytometer (Beckton Dickinson, San Jose, CA) and analyzed with FlowJo software (Treestar, Ashland, OR). Antibody responses to neoantigen sheep red blood cells (SRBC) were determined at 148 to 248 days after transplantation [26,27].

Statistical analysis

The log-rank test was used for comparison of survival and time to count recovery between two groups. Wilcoxon rank sum test was used for comparison of transfusion support. McNemar's test was used for evaluating the cell viability and dominant unit engraftment outcome.

Results

DLA-identical sibling cord blood transplantation

For the initial experiments, cord blood units were infused into DLA-identical siblings from a prior litter of the same breeding pair. Each unit contained 0.3 – 0.8×10⁷ TNC/kg. Because of the low total leukocyte dose in each unit, we combined two to four DLA-identical units for each CBT. Table 1 summarizes the results. Five dogs received 2–4 units of DLA-identical cord blood grafts following 9.2 Gy TBI. Four dogs engrafted and survived long-term with stable donor engraftment and evidence of robust immune reconstitution (Figure S.3). One dog was euthanized on day +13 due to sepsis. None of the dogs developed GVHD. The median time to engraftment with sustained ANC>500/µL was 28 (range, 21–32) days. The median time to engraftment with sustained platelets> 50,000/µL was 58 (range, 35–99) days. Mixed hematopoietic chimerism was initially observed in all recipient dogs, although one dominant unit emerged that eventually exceeded >80% PBMC and >90% granulocyte chimerism.

Table 1.

DLA-identical sibling multiple unit cord blood transplantation

Recipient Dog #	Donor Unit #	DLA- DRB1 alleles	TNC (×107/kg)	CD34 (×105/kg)	CD3 (×106/kg)	% CD34+ viable	% PBMC chimerism at 1 year	Transplant Outcome
G567	G464 u.5	00401/	0.6	0.6	1.3	ND	92	Alive > 3 yr.
	G464 u.9	01301	0.6	0.4	1.9		6
G568	G464 u.1	01301/	0.4	0.3	0.5	69	NE	ET day 13
	G464 u.2	01501	0.3	0.2	1.2	75		Sepsis, E. coli
	G464 u.6		0.4	0.4	0.7	70
	G464 u.10		0.3	0.2	0.5	71
G569	G464 u.4	00401/	0.6	0.5	2.6	ND	7	Alive > 3 yr.
	G464 u.11	01901	0.4	0.4	1.2		89
	G464 u.12		0.4	0.4	0.8		3
G570	G464 u.3	01501/	0.6	0.5	1.6	95	99	Alive > 3 yr.
	G464 u.7	01901	0.4	0.4	1.0	83	0
G545	E333 u.1	00801/	0.5	0.4	0.9	98	80	Alive > 3 yr.
	E333 u.2	01401	0.6	0.6	1.6	86	20

The donor unit is identified by maternal dog number followed by the cord/fetal unit in the order of collection at the time of cesarean section. Dog leukocyte antigen (DLA) alleles indicate nucleotide sequence nomenclature for DLA-DRB1 (MHC class II) or DLA-88 (MHC class I) (http://www.ebi.ac.uk/ipd/mhc/). The cord blood units were DLA-identical with each other and the recipient. Boldface indicates dominant engrafting cord blood unit. TNC = total nucleated cells, kg = kilogram (recipient weight). There was significant unit-to-unit variability in the percent nucleated red blood cells (nRBC). Therefore, the TNC shown excludes all nRBC. PBMC = peripheral blood mononuclear cell. ET= euthanized due to meeting defined criteria of poor condition (indicated by day after transplantation and cause). ND= not done. yr.= years. The % viable CD34+ cells were determined by flow cytometry with an aliquot from the unit after thawing. NE= not evaluable.

Next we asked if infusion of one DLA-identical cord blood unit with a low cell dose could engraft and promote survival after 9.2 Gy TBI. Three dogs underwent single unit CBT. Two dogs were euthanized early due to sepsis, on day +15 and 17, respectively, with findings of marrow aplasia and hemorrhage at necropsy (Table 2). The one evaluable recipient engrafted with sustained ANC>500/µL on day 34 and platelets >50,000/µL on day 63. This surviving recipient developed >98% donor PBMC chimerism by one year and had successful immune reconstitution (Figure S.3). These initial results were not definitive, but suggested that the infusion of a larger cell dose or multiple units of DLA-identical cord blood resulted in a greater likelihood of sustained engraftment and survival.

Table 2.

DLA-identical sibling single unit cord blood transplantation

Recipient Dog #	Donor Unit #	DLA- DRB1 alleles	TNC (×107/kg)	CD34 (×105/kg)	CD3 (×106/kg)	% PBMC Chimerism at 1 year	Transplant Outcome
G730	G692 u.8	00201/	0.8	0.8	1.4	98	Alive > 2.5 yr.
		00102
G732	G692 u.7	00201/	0.4	0.4	1.0	NE	ET day 15
		01501					Sepsis
G734	G692 u.9	00201/	0.4	0.3	0.7	NE	ET day 17
		00102					Sepsis M. morganii

The donor unit is identified by maternal dog number followed by the cord/fetal unit in the order of collection at the time of cesarean section. Dog leukocyte antigen (DLA) alleles indicate nucleotide sequence nomenclature for DLA-DRB1 (MHC class II) or DLA-88 (MHC class I) (http://www.ebi.ac.uk/ipd/mhc/). The cord blood units were DLA-identical with the recipient. TNC= total nucleated cells, kg= kilogram (recipient weight). There was significant unit-to-unit variability in the percent nucleated red blood cells (nRBC). Therefore, the TNC shown excludes all nRBC. PBMC = peripheral blood mononuclear cell. ET= euthanized due to meeting defined criteria of poor condition (indicated by day after transplantation and cause). yr.= years. NE= not evaluable.

Unrelated Canine Cord Blood Bank

In order to develop a dog model that more closely addressed the clinical problem of single unit low cell dose unrelated CBT for adult human patients, we next focused on the development of a cord blood bank with common Class II alleles. After breeding a total of 30 dog pairs, there were 187 cord blood units collected and cryopreserved, representing 15 different combinations of the two most polymorphic MHC-class I and II alleles, DLA-88 and DRB1, respectively. Next, we identified unrelated transplant recipients that were DLA-matched for all DRB1 and DLA-88 alleles. The recipients were obtained from a separate, genetically diverse, outbred colony.

DLA-matched unrelated cord blood transplantation

For the DLA-matched unrelated CBT model, we first infused two cord blood units into recipients. Table 3 summarizes the results. Five dogs received double unit DLA-matched unrelated CBT following 9.2 Gy TBI. Each unit contained 0.4 – 1.7×10⁷ TNC/kg. All 5 dogs engrafted, but one dog was euthanized on day 50 due to acute pneumonitis. None of the dogs developed GVHD. The median time to engraftment with sustained ANC>500/µL was 17 (range, 16–19) days (Figure 1A). The median time to engraftment with sustained platelets >50,000/µL was 35 (range, 29–38) days. Mixed hematopoietic chimerism was initially observed in all recipient dogs, although there emerged one dominant unit that exceeded >80% PBMC and >90% granulocyte chimerism by day 35 to 133 after CBT (Figure 2). The TNC, CD34, CD3, CD4 and CD8 cell dose, sex mismatch and order of infusion did not predict the dominant engrafting unit. However, combining the available data from both related and unrelated double unit CBT recipients, the viability of the CD34+ cells at the time of thawing and infusion of the two units appeared to predict the engrafting unit. Among the 7 evaluable recipients (Table 1 and Table 3 and Figure 2), the unit with the greatest CD34+ viability at the time of infusion was subsequently the dominant engrafting unit, p = 0.03.

Table 3.

Double unit DLA-matched, unrelated cord blood transplantation

Recipient			Cord Blood Unit					Transplant Outcome
	DLA alleles		Donor Unit #	Cell Dose			% CD34+ viable	% PBMC chimerism at 1 year	Graft status
Dog #	DRB1	Class I		TNC (×107/kg)	CD34 (×105/kg)	CD3 (×106/kg)
G877	00102/	01201/	G879 u.5	0.5	0.5	1.4	98	96	Engrafted,
	01501	50201	G833 u.4	1.7	1.1	4.1	77	0	alive >2 yr.
G878	00201/	01201/	G756 u.3	0.8	0.4	1.9	91	6	Engrafted,
	01501	50801	G755 u.3	1.1	0.7	3.5	97	91	alive >2 yr.
G903	00201/	50201/	G755 u.2	1.2	0.6	3.8	90	94	Engrafted,
	00102	50801	G755 u.4	1.4	1.0	4.5	85	0	alive >2 yr.
G905	00102/	01201/	G779 u.3	0.9	1.1	3.0	92	N/A	Engrafted
	01501	50201	G833 u.8	1.3	0.8	2.7	87	N/A	ET day 50 pneumonitis
G910	00102/	01201/	G879 u.1	0.4	0.3	1.2	85	10	Engrafted,
	01501	50201	G881 u.4	1.4	0.8	1.7	85	90	alive >2 yr.

The donor unit is identified by maternal dog number followed by the cord/fetal unit in the order of collection at the time of cesarean section. Dog leukocyte antigen (DLA) alleles indicate nucleotide sequence nomenclature for DLA-DRB1 (MHC class II) or DLA-88 (MHC class I) (http://www.ebi.ac.uk/ipd/mhc/). The donor units were DLA-matched (for DLA-DRB1 and DLA-88) with each other and the recipient. Boldface indicates dominant engrafting cord blood unit. TNC= total nucleated cells, kg= kilogram (recipient weight). The TNC shown excludes all nucleated red blood cells. The % viable CD34+ cells were determined by flow cytometry with an aliquot from the unit after thawing. PBMC = peripheral blood mononuclear cell. ET= euthanized due to meeting defined criteria of poor condition (indicated by day after transplantation and cause). yr.= years. N/A = not applicable.

Figure 1.

The absolute neutrophil count (ANC) and platelet count per microliter (log₁₀ scale) from dogs receiving unrelated cord blood transplantation (CBT) on day 0 after 9.2 Gy TBI. A: Double-unit CBT group. B: Single-unit CBT group.

Each line/symbol represents blood counts from individual dogs. ET= euthanized due to meeting defined criteria of poor condition after transplantation (for details, see Table 3 and Table 4).

Figure 2.

Serial chimerism analyses from peripheral blood mononuclear cells for five dogs that received double unit unrelated cord blood transplantation. The first infused donor unit had the lower total cell dose. Each cord blood unit cell dose is shown in parenthesis (TNC × 10⁷/kg). Granulocyte chimerism for the dominant unit was consistently ≥ % PBMC chimerism (data not shown).

All surviving dogs had immune reconstitution with recovery of T cell function, measured by mixed lymphocyte reaction assay (data not shown), antibody response to sheep red blood cells and T cell subset recovery to pre-transplant levels developing between 150 to 250 days after CBT (Figure 3).

Figure 3.

(A) Immune recovery of T- and B-cell subsets in 4 dogs after double unit unrelated cord blood transplantation (CBT). Shaded region indicates normal range of values (median ± 1 standard deviation from 20 untreated dogs). (B) Primary and secondary antibody titer response against the neoantigen sheep red blood cells injected on day +148-184 and +218-248 after CBT, respectively.

Next, we asked if infusion of one unit of low cell dose unrelated cord blood could achieve engraftment and survival. Five dogs received one unit of DLA-matched unrelated cord blood following 9.2 Gy TBI (Table 4 and Figure 1B). Each unit contained 0.6 – 0.9×10⁷ TNC/kg. One dog was euthanized because of pneumonitis on day 11 and was not evaluable for engraftment. Four dogs achieved first day of ANC>500/µL at a median of 24 (range, 21 to 28) days after CBT. However two dogs had subsequent graft rejection, remained neutropenic, then developed bacterial sepsis and were euthanized on days 70 and 74, respectively. One dog had neutrophil engraftment but remained platelet transfusion dependent and developed pneumonitis and was euthanized on day 54. Only one dog had neutrophil and platelet engraftment with recovery on day 21 and 68, respectively, but developed sepsis and was euthanized on day 105. In summary, all 5 recipients of single unit CBT died due to infectious complications, despite intensive supportive care and prompt intervention with broad-spectrum antibiotics.

Table 4.

Single unit DLA-matched, unrelated cord blood transplantation

Recipient			Cord Blood Unit				Transplant Outcome
	DLA alleles			Cell dose			Day of euthanasia, graft status, necropsy findings
Dog #	DRB1	Class I	Donor Unit	TNC (×107/kg)	CD34 (×105/kg)	CD3 (×106/kg)
G940	00102/	01201/	G881 u.5	0.9	0.5	2.0	ET day 74, Graft rejection.
	01501	50201					Sepsis: enterococcus sp., staphylococcus
G944	00102/	50801/	H021 u.4	0.6	0.7	0.7	ET day 11, Graft NE.
	00201	50201					Septic pneumonitis, multiple organisms
H160	00102/	01201/	G881 u.6	0.8	0.6	1.0	ET day 105, Engrafted.
	01501	50201					Sepsis: E. coli, hemorrhage
H162	00102/	01201/	H019 u.9	0.6	0.6	0.7	ET day 54, Neutrophil graft.
	01501	50201					Pneumonitis, sepsis: enterococcus sp.
H164	00102/	50801/	H021 u.2	0.6	0.8	1.1	ET day 70, Graft rejection.
	00201	50201					Sepsis: Pseudomonas, enterococcus sp.

The donor unit is identified by maternal dog number followed by the cord/fetal unit in the order of collection at the time of cesarean section. Dog leukocyte antigen (DLA) alleles indicate nucleotide sequence nomenclature for DLA-DRB1 (MHC class II) or DLA-88 (MHC class I) (http://www.ebi.ac.uk/ipd/mhc/). The donor units were DLA-matched with the recipient. TNC= total nucleated cells, kg= kilogram (recipient weight). The TNC shown excludes all nucleated red blood cells. ET= euthanized due to meeting defined criteria of poor condition (indicated by day after transplantation and cause).

Compared to unrelated double unit CBT, single unit CBT was associated with significantly worse survival (p = 0.03). The times to neutrophil and platelet recovery were markedly delayed in single unit recipients, p = 0.005 and 0.005, respectively. The differences in supportive care given to the two cohorts were also significant. The median number of blood transfusions given to unrelated double vs. single cord blood unit recipients was 4 vs. 17, respectively, p = 0.01.

Discussion

We have developed a large animal model of CBT. The initial studies were performed with DLA-identical sibling cord blood units to develop the methods and establish the feasibility of CBT in the dog model. We showed that transplantation of multiple cord blood units rescued dogs from myeloablative TBI, and that although all the units were DLA-identical with each other and the recipient, one unit eventually became the dominant source of hematopoiesis. In contrast, only 1 of the 3 recipients of a single DLA-identical littermate cord blood unit survived. However the prolonged time to engraftment and early infectious complications and the limited availability of DLA-identical littermate cord blood units precluded definitive comparison between single and multiple units of CBT.

In order to more accurately model the clinical setting of low cell dose unrelated CBT for adult human patients, we next developed a dog cord blood bank with non-homozygous combinations of the six most common DLA-DRB1 alleles. The allelic sequences of DLA-DRB1 and DLA-88, the most polymorphic canine MHC class II and class I alleles, respectively, were determined for each collected cord blood unit, and unrelated recipients matched for both DRB1 and DLA-88 alleles were identified.

In this experiment, cord blood units with low cell dose were used to model the clinical situation in which the best matched unrelated cord blood unit was well below the standard TNC/kg doses generally accepted for clinical CBT. With double unit unrelated CBT, all five recipients engrafted and one unit contributed >80% of myeloid and lymphoid hematopoiesis within 4 to 16 weeks after CBT.

Perhaps because of the limited number of animals studied, we were unable to identify if greater TNC, CD34 or T cell dose predicted which unit became the dominant contributor to engraftment. However, in accordance with a recent report by Scaradavou et al. [13], we observed that among the evaluable recipients of double unit CBT, the unit with the greatest CD34+ cell percent viability at the time of thawing and infusion predicted the subsequent dominant engrafting unit. Another emerging concept is that there is CD8 T cell mediated competition between the two donor units which determines the dominant engrafting unit [28]. Further studies in this CBT model will be needed to understand the engraftment process, although we can speculate that the CD34+ viability may reflect the overall health of the thawed units, and the immune mediated competition between units favors the unit with the greater cellular viability.

After double unit CBT immune reconstitution was relatively prompt, with various measures of T cell recovery returning to pre-transplant levels within 150 to 250 days, and subsequently remaining stable. In contrast, despite optimal intensive supportive care, single unit DLA-matched unrelated CBT did not achieve reliable engraftment and all five dogs died within 105 days after transplantation. The time to engraftment of neutrophils and platelets after single unit CBT was significantly delayed compared to double unit CBT. Similarly, even though neutrophil engraftment occurred in two recipients, immune recovery was inadequate to help protect against bacterial pathogens despite very close clinical monitoring and prompt, aggressive treatment with broad spectrum antibiotics. Although the prolonged duration of CSP/MMF may have contributed to the poor outcome of the single unit CBT recipients, data from other experiments suggested this was unlikely. In previous studies after DLA-identical hematopoietic cell transplantation we showed that extending post-grafting CSP immunosuppression successfully prevented graft rejection and did not increase infectious complications [29]. In another cohort of recipient dogs given DLA-mismatched CBT, prolongation of CSP/MMF prevented graft rejection, improved survival, and did not increase the risk of infectious death (manuscript in preparation).

These results established a minimum threshold cell dose required for engraftment and survival after unrelated CBT in this dog model which was within the range 1.0 –1.8 ×10⁷ TNC/kg. The minimum threshold cell dose for CBT engraftment in DLA-identical siblings appears to be slightly lower (<0.8×10⁷ TNC/kg). This may be due to the greater genetic disparity between unrelated donors and recipients. Due to the cost limitations of the experimental large animal model, we did not test if a single unit CBT above the minimum threshold cell dose was sufficient for engraftment and survival. We showed that after double unit CBT, the unit that became the dominant engrafting unit had a cell dose that would have been predicted to be insufficient to achieve engraftment and survival if given as a single unit. Comparing double versus single unit CBT, it appears that the non-engrafting unit enhanced the engraftment of the dominant unit with more rapid recovery of neutrophils and platelets and a lower rate of graft rejection. The data suggest that the non-engrafting unit had an effect equivalent to increasing the total cell dose of the engrafting unit.

Our results in this model are very relevant to the clinical experience with CBT. Clinical use of CBT for adults was limited for several years because transplantation of single unit, low cell dose cord blood was associated with high transplant related mortality ranging from 40% to 68% [11,30,31]. There are relatively few single units with a sufficiently high cell dose available for adults [8]. Similar to the findings of our study, the introduction and success of double-unit CBT in non-randomized phase II clinical trials resulted in an increased number of adult patients eligible for transplantation since the combination of two units, each with a low cell dose, reached the safer threshold dose of 3×10⁷ TNC/kg [5–8,10]. A randomized clinical CBT study directly comparing one versus two low cell dose units is not feasible, due to the concern of high transplant related mortality in low cell dose single unit recipients. Instead, the results of our study strongly support the use of increased cell dose achieved with double unit CBT.

The significance of this CBT model may be enhanced as we address other important issues facing clinical CBT. Strategies to improve engraftment and survival after partially MHC-mismatched CBT can now be directly studied in dogs in order to reduce the risk for patients participating in clinical studies. This model allows us to test if one (or more) MHC-mismatched unit(s) can be given to promote engraftment of a single MHC-matched or partially matched low cell dose unit. Additional questions suited for study in this double unit CBT model include how to ensure the engraftment of the best MHC-matched unit and how to prevent an MHC-mismatched unit from becoming dominant and causing GVHD. Other approaches that can be rigorously tested in this model include the infusion of ex vivo expanded cord blood cells to facilitate engraftment of low cell dose units and treatment to enhance the homing of stem cells to the bone marrow.

In addition to the relevance of this model for clinical CBT, another significant impetus for this project came from the need to develop better treatment protocols for victims of radiation accidents or nuclear terrorist attacks. Cord blood is a stockpiled, readily available “off the shelf” resource that can be rapidly used in the setting of accidental high dose TBI exposure (ranging from 7 to 11 Gy). For radiation victims with the hematopoietic syndrome and low probability of endogenous hematopoietic recovery, CBT is an important therapeutic treatment option. This dog CBT model will allow us to address several critical questions including: What additional conditioning is needed with increasing time delay between radiation exposure and CBT? What is the optimal post-grafting immunosuppression? What are the acceptable parameters of MHC mismatch for supplemental cord blood units when the cell dose of the best-matched cord blood unit is below recommended levels? Most of these questions are also relevant to the further clinical development of CBT for patients with malignant and non-malignant hematologic diseases.