Optimization of Intrabone Delivery of Hematopoietic Progenitor Cells in a Swine Model Using Cell Radiolabeling with [89]zirconium

Abstract

Intrabone (IB) hematopoietic cell transplantation (HCT) of umbilical cord blood in humans remains experimental and the technique has not been optimized. It is unknown whether hematopoietic progenitor cells (HPCs) injected IB are initially retained in the marrow or rapidly enter into the venous circulation before homing to the marrow. To develop an IB-injection technique that maximizes HPC marrow-retention, we tracked radiolabeled human HPCs following IB-injection into swine. We developed a method to radionuclide-label HPCs using a long-lived positron emitter ⁸⁹Zr and protamine sulfate that resulted in cellular-retention of low-dose radioactivity. This approach achieved radioactivity levels sufficient for detection by positron emission tomography with both high sensitivity and spatial resolution when fused with computed tomography. We found that conditions utilized in pilot IB-HCT clinical trials conducted by others led to both rapid drainage into the central venous circulation and cellular extravasation into surrounding muscle and soft tissues. By optimizing the needle design, using continuous real-time intra-marrow pressure monitoring, and by reducing the infusion-volume and infusion-rate, we overcame this limitation and achieved high retention of HPCs in the marrow. This method of IB cellular delivery is readily applicable in the clinic and could be utilized in future investigational IB-HCT trials aimed at maximizing marrow retention of HPCs.

Introduction

Although UCB HCT can cure patients with advanced hematological malignancies, graft failure remains a major cause of mortality that impedes the wider applicability of this transplant approach in adults. Recent preclinical animal data and pilot clinical trials suggest IB-HCT may improve engraftment in adult recipients of UCB HCT compared to those receiving conventional IV injection of UCB. Impediments to the progress of IB-HCT include uncertainty regarding injection techniques that optimize the retention of HPCs in the BM. Currently it remains unknown whether IB-injected HPCs in humans are initially retained in the BM or rather rapidly enter into venous circulation before homing back to the BM. Surprisingly, pilot trials in humans were initiated before preclinical studies were developed to optimize techniques of IB-HCT. Preclinical studies require labeling HPCs in a fashion that would allow them to be tracked in vivo following IB-injection to provide information necessary to develop an injection technique that maximizes cellular-retention in the BM.

Although ¹¹¹In oxine traditionally has been used to label HPCs, this approach exposes cells to high doses of radiation which may limit their viability (1) and function (2) in vivo. Moreover, images are suboptimal because ¹¹¹In is a single-photon emitter detected with SPECT cameras, which have limited resolution and sensitivity. An alternative to single-photon emitters are positron-emitters that emit two oppositely directed γ-rays after annihilation with an electron. PET has inherently higher sensitivity (up to 10-fold) and better spatial resolution than SPECT enabling lower doses of radioactivity to be used without compromising its quantitative capacity (3). In this study, we applied PET cell-imaging techniques using the long-lived PET emitter⁸⁹Zr (Aras, O, Berg, M, Pantin, J, Szajek, L, Paik, C, Frank, J, Sato, N, Childs, R, Choyke, P, Labeling of NK Cells with 89Zr for Positron-Emission Tomography (PET) Imaging: A New Method for in Vivo Cell Tracking. Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012, Chicago IL.http://archive.rsna.org/2012/12037076.html Accessed January 8, 2015), to visualize HPCs in vivo that were transplanted via IB-injection. Using ⁸⁹Zr-labeled HPCs, we aimed to optimize the IB-HCT procedure to retain HPCs injected into BM of domestic swine, a large animal whose pelvic size and anatomy share features with humans.

Methods

Anesthesia, monitoring, and porcine pelvis IB access

Animal procedures were approved by the institutional Animal Care and Use Committee (Study H-0233). Domestic swine (Sus scrofa domesticus) of both genders weighing between 38 and 70 kg were utilized. All experiments used general anesthesia; animals were sedated with ketamine, midazolam, and propofol prior to endotracheal intubation and general anesthesia was maintained with isoflurane. Analgesia was provided with local bupivacaine and IV buprenorphine. Animals had continuous EKG (Physio-Control LifePak20), end-tidal CO₂ (Ohmeda 5250 RGM), oral temperature, noninvasive pulse oximetry (NeoVet, Centaurus Medical, Saint Petersburg, FL), and intra-arterial blood pressure monitoring (ADI Powerlab with Quad Bridge Amp, ADInstruments, Colorado Springs, CO) throughout each experiment. Intravenous vecuronium was used for paralysis during imaging. Internal jugular vein and internal carotid artery cannulation were performed after cervical cut-down. A Swan-Ganz catheter (Edwards Lifesciences Corporation, Irvine, CA) was inserted via the internal jugular vein under fluoroscopic guidance, to allow continuous measuring of pulmonary artery pressures. For IB-injection, the posterior–superior iliac crest was exposed after minimal dissection. A truncated Jamshidi BM needle (Cardinal Health, Dublin, OH) with a custom threaded base was used to establish IB-access in animals for direct injection of iodinated contrast media. An OnControl™ driver (Vidacare Corp., San Antonio, TX) was used to insert modified needles having distal fenestrations (Figure 1A) for HPC IB-injections. A hemostasis screw type Y-connector (Cordis Corporation) was attached to the inserted BM needle, allowing passage of a 2Fr Millar Mikro-Tip^® pressure catheter (ADInstruments) into the BM in order to provide real-time measurement of injection-pressures (Figure 1B). Intra-marrow cavity-pressure was measured continuously during IB-injection and was compared to systemic arterial pressure, pulmonary arterial pressure, and electrocardiography on a PowerLab data acquisition system (ADInstruments). Acquired hemodynamic data were analyzed using LabChart^®7 (ADInstruments). At the conclusion of each experiment, animals were euthanized under anesthesia with intravenous KCl.

Figure 1:

(A) Shortened OnControl™ needle with distal fenestrations (arrows) added. In all experiments, this needle was driven directly into the iliac-crest to the hub. A longer needle was used for CT-guided percutaneous access. (B) Simultaneous infusion and intra-marrow pressure monitoring with a Millar catheter through a hemostasis screw type Y-connector connected to an OnControl™ needled inserted into the iliac-crest.

Porcine pelvic vascular imaging and hemodynamic monitoring

To evaluate flow through the iliac-BM, abdominal vessels were identified using iopamidol-370 contrast injected into the distal abdominal aorta with a 5Fr Glidecath (Terumo Medical Corporation, Somerset, NJ) inserted via the internal carotid artery. To evaluate the anatomy of the veins, iopamidol-370 contrast was IB-injected under anesthesia (boluses of 5 or 10mL at rates of 0.1–2.5 mL/s) while scanning with 320-row detector CT (Aquilion ONE™, Toshiba Medical, Tustin, CA). Pulmonary artery, carotid artery and intra-marrow pressures were monitored and recorded continuously both during and after injection of contrast media and HPCs. In select animals, vascular occluders (In Vivo Metric, Healdsburg, CA) were placed around the terminal branches of the abdominal aorta to selectively control flow through the internal or external iliac arteries, and around the iliac veins to control venous drainage from the pelvis. Selective catheterization of internal or external iliac arteries, with selective vessel occlusion, was performed to determine the dominant vessels supplying blood to the iliac-BM.

Cell-labeling

HPCs utilized in all experiments had been collected by apheresis from healthy human subjects following 5 days of daily G-CSF and one dose of plerixafor after informed consent and enrollment on a National Heart Lung and Blood Institute Institutional Review Board approved protocol (ClinicalTrials.gov Identifier: NCT00082329), then were CD133+ (4) selected using the Miltenyi CliniMacs system and were cryopreserved viably.

Zr was obtained from the local cyclotron facility utilizing the nuclear reaction Y (p, 2n) ⁸⁹Zr and an in-house GE PETtrace™ beam-line (GE Healthcare, Chalfont St. Giles, United Kingdom) as described previously (Sato et al submitted). ⁸⁹Zr was provided in 1 M oxalic acid in greater than 96% radiochemical yield. Prior to cell-labeling, ⁸⁹Zr-oxalate was neutralized with 2 M sodium bicarbonate and 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid, then adjusted to 1 mL by water.⁸⁹Zr cell-labeling was performed as previously described (Sato et al submitted). Briefly, frozen HPCs were thawed and incubated in the mixture of neutralized ⁸⁹Zr (0.05–0.1mCi), protamine (40 μg/mL), and heparin (2 IU/mL) in phosphate-buffered saline (PBS) for 30 min at 37°C. The cells were washed with PBS twice and then with reconstituted Df (Desferal^®, Novartis Pharmaceuticals, Basel, Switzerland) in PBS in order to remove the unbound ⁸⁹Zr. The amount of cellular radioactivity before and after the labeling procedure was routinely determined with a WIZARD2 automatic γ-counter (Perkin Elmer, Waltham, MA) and dose calibrator (Capintec, Ramsey, NJ). Cell viability was assessed immediately after the labeling procedure using trypan blue then cells were injected immediately into animals. In one experiment, cell survival after labeling with ⁸⁹Zr was assessed in vitro for up to 14 h.

To determine the retention of radioisotope within the HPCs, 1×10^{6 89}Zr labeled human CD133+ cells (⁸⁹Zr-hCD133+) were cultured in RPMI medium and ⁸⁹Zr-activity in cells and medium was counted by a γ-counter at various time-points after labeling.

HPC IB-injection and in vivo PET/CT imaging

To evaluate the in vivo distribution of non-cell associated ⁸⁹Zr, ⁸⁹Zr-Df chelate (1 mCi in 10mL solution) was IV-injected into pigs followed immediately by PET/CT imaging. To determine the in vivo distribution of human HPCs, all animals received 2×10^{8 89}Zr-hCD133+ in 10mL normal saline (8–1 μCi total radioactivity) which was injected either IV into the internal jugular vein, IB into the swine iliac crest, or intra-arterially by selective catheterization of the internal iliac artery, followed by PET/CT imaging. Animals were heparinized to an ACT of 300 s or greater and received a Df IV-infusion (5 mg/kg/h) to chelate any free⁸⁹Zr that may have been released from any fragmented or dead cells.

Hand-injections of ⁸⁹Zr-hCD133+ into pigs were delivered using the minimal force needed to advance the plunger. For the non-hand-injection cohorts, IB-injections were delivered at a controlled-infusion rate of 0.2 mL/min using a Harvard PHD 22/2000 Advance Syringe Pump (Harvard Apparatus, Holliston, MA).

PET scans were performed using a Gemini TF clinical PET/CT scanner (Philips Healthcare, Andover, MA). Following injection, animals were imaged in 3D time-of-flight mode with a spatial resolution of 4.8mm at the center of the field of view. Images were reconstructed using the default row-action maximum-likelihood algorithm iterative reconstruction, with standard corrections for random, scatter, attenuation, and normalization.

Image analysis

Three VOIs were manually drawn within the swine PET/CT scans following capture: one VOI enclosing the iliac-BM volume near the injection-site (2) another VOI enclosing the lung volume (3) and a final VOI enclosing areas outside the iliac-BM and lung volumes. The sum of measured activities of the three volumes estimates the total injected radioactivity into the swine. The %ID within each VOI was assessed by dividing each VOI by the sum of the three VOIs × 100.

Results

Porcine pelvic vascular supply and drainage following IB-injection

Utilizing selective catheterization and 3D CT angiography, the porcine iliac-BM vascular supply and drainage following IB-injection was determined. The porcine external and internal iliac arteries arise as terminal branches of the abdominal aorta rather than from a common iliac trunk (Figure S1) however, the distribution to the pelvis is similar to humans. Blood perfusion to the iliac-BM is supplied by the internal iliac artery. Venous drainage of the iliac-BM occurs via short veins that eventually drain into a common iliac vein. Following IB-injection, contrast was observed to move first into the iliac-BM followed by opacification of short pelvic veins then the common iliac vein (Video S1). When the common iliac vein was occluded, contrast drainage from the iliac-BM occurred via veins along the iliac crest into the lumbar venous plexus, and via crosspelvic collaterals (and also via uterine veins in female pigs) to the contralateral common iliac vein.

Contrast IB-injection into the BM at rates exceeding 0.2 mL/s, volumes >10mL and IB-injections performed by hand, all resulted in contrast leakage at the site of injection outside the iliac-BM. When more than one IB access site was made in the ipsilateral iliac crest, contrast was observed to egress from the non-injected site regardless of the infusion rate and the distance between the two sites (Figure S2 and Video S2).

Hemodynamic changes during IB-injection

Reducing injection-volumes and slowing the injection-rate both reduced intra-marrow pressures following IB-injection. In particular, by not exceeding injection-rates faster than 0.2 mL/s or injection-volumes of>10 mL, intra-marrow pressures typically increased only slightly above systemic arterial pressures and were accompanied by a minimal increase in post-injection peak pulmonary artery systolic pressures (PASP) (Table 1). However, larger injection-volumes and more rapid injection-rates were associated with markedly elevated intra-marrow pressures and substantial increases in post-injection PASP (Table 1 and Figure 2). At larger volumes and more rapid injection-rates, the change in post-injection PASP appeared to decrease over time, however because increasing volume and rate injections were performed sequentially in each animal, baseline PASP remained elevated whenever injections exceeding rates of 1 mL/s were performed resulting in lower net difference. After high-volume/high-rate IB-injections, all animals developed systemic hypotension, tachycardia, hypoxemia and increased PASP suggesting pulmonary embolism had occurred. The occurrence of pulmonary embolism was confirmed in several swine developing these physiologic findings by subsequent CT or fluoroscopic angiography (data not shown).

Table 1:

Peak intra-marrow pressure and change in pulmonary artery systolic pressure following IB-injection at varying rates and volumes

Volume injected (mL)	Rate injected (mL/s)	Number of experiments	Δ from baseline in intra-marrow pressure (mmHg), mean ± SE	Δ from baseline in peak pulmonary artery systolic pressure (PASP) (mmHg), mean ± SE
10	0.1	4	159.3 ± 57.7	1.5 ± 0.7
10	0.2	3	300.3 ± 149.3	3.5 ± 1.0
10	0.5	2	803.4 ± 483.0	13.0 ± 5.1
10	1.0	2	1196.0 ± 265.5	8.9 ± 9.3
10	2.5	3	1689.0 ± 116.0	2.2 ± 1.0
5	0.1	2	128.7 ± 60.2	−0.5 ± 0.4
5	0.2	2	256.7 ± 42.7	1.3 ± 1.3
5	0.5	1	802.9	6.1
5	1.0	1	1490.8	0
5	2.5	1	997.5	2

Figure 2: Physiologic parameters following IB-injection of 10 mL of iopamidol-370 contrast at 1.0 mL/s (rapid) and at 0.1 mL/s (slow) in a representative animal experiment.

With rapid injection, there was a significant peak in intra-marrow pressure followed by an increase in PA pressure, a decrease in systemic arterial pressure, and the development of tachycardia. When injection-rates were slowed, changes in these parameters were not observed.

Cell-labeling

Thawed hCD133+ cells were labeled with ⁸⁹Zr up to 7.68 μCi/million cells. At this dose, in vitro studies showed the majority of hCD133+ cells remained viable 14 h after labeling, establishing that the labeling process did not affect hCD133+ cell survival in vivo. Gradual release of ⁸⁹Zr from labeled-cells was observed to occur over time (Figure 3), which could protect labeled-cells in vivo from potentially detrimental effects of protracted radiation exposure.

Figure 3: hCD133₊ cells were labeled with neutralized ⁸⁹Zr as described.

(A) ⁸⁹Zr labeling of hCD133+ cells did not affect their survival in vivo for up to 14 h. (B) ⁸⁹Zr was gradually released from ⁸⁹Zr-hCD133+ cells.

Injection of ⁸⁹Zr-Df IV or IB

Following IV- or IB-injection of ⁸⁹Zr-Df (non-cell bound chelate), measurements were obtained of the radioactivity in and external to the iliac-BM, and lungs including the VOI maximum and mean pixel-value, VOI in milliliter, totalradioactivity in counts, and %ID after the first hour. One hour following IV or IB ⁸⁹Zr-Df administration without cells, the agent was largely confined to the blood pool, renal collecting system, ureters and bladder with little or no lung radioactivity observed. In the case of IB-injections, some radioactivity was observed to remain at the site of injection (33.5%ID) (Figure 4A and Table 2). There was no retention of ⁸⁹Zr in the lungs when ⁸⁹Zr-Df without cells was administered.

Figure 4:

Fusion PET/CT coronal images at 1 h following (A) IB-injection of ⁸⁹Zr-Df with radioactivity in the renal pelvises and bladder (Pig ID# 32890); (B) IV-injection of ⁸⁹Zr-hCD133+ with radioactivity confined to the lung (Pig ID#32891); (C) Hand IB-injection of ⁸⁹Zr-hCD133+ into two sites on the right iliac-crest demonstrating radioactivity leaked at the first site of injection (arrow) and radioactivity within the iliac-BM and lungs (Pig ID#34489); (D) hand IB-injection of ⁸⁹Zr-hCD133+ into a single site on the right iliac-crest demonstrating radioactivity leaked at the sight of injection (arrow) and radioactivity within the iliac-BM and lungs (Pig ID# 31717); (E) slow non-manual IB-infusion (0.2 mL/min) of ⁸⁹Zr-hCD133+ into the right iliac-crest demonstrating radioactivity confined within the iliac-BM (Pig ID# 33189). Note that in (D) there is radioactivity highlighted in syringes of ⁸⁹Zr-oxalate placed to the left of the swine for calibration and decay correction (star).

Table 2:

The region of interest (ROI) maximum and mean pixel value, ROI volume, total radioactivity in counts, and percent of injected dose (%ID) after the first hour for within the iliac marrow (intra-iliac-BM); outside the iliac-BM but within the pelvis (extra-pelvic); and in the lungs; for each pig administered ⁸⁹Zr-deferoxamine chelate (⁸⁹Zr-Df) or ⁸⁹Zr-radiolabeled human CD133+ cells (⁸⁹Zr-hCD133+) either IV or IB

Pig ID#	Weight (kg)	Injection	Group	Volume injected (mL)	Intra-iliac marrow VOI (max.)	Intra-iliac marrow VOI (mean)	Volume of intra-iliac marrow VOI (mL)	Extra-pelvic VOI (max.)	Extra-pelvic VOI (Mean)	Volume of Extra-pelvic VOI (mL)	Lung VOI (max.)	Lung VOI (mean)	Volume of lung VOI (mL)	Intra-iliac marrow total radioactivity	Extra-pelvic total radioactivity	Lung total radioactivity	Intra-iliac marrow %ID	Extra-pelvic %ID	Lung %ID
32890	52	89Zr-Df	Control	10	338.9	114.42	9.49	91.45	16.32	132.26	2.39	0	2159.22	1085.85	2158.48	0	33.5%	66.5%	0.0%
32891	53	89Zr-hCD133+	IV	10	0	0	0	1.98	0.17	27.82	17.21	0.38	3556.15	0	4.73	1351.34	0.0%	0.3%	99.7%
34489	43	89Zr-hCD133+	IB-hand (2 ipsilateral sites)	5 (×2)	503.83	174.36	5	618.99	225.93	8.03	38.04	1.28	1693.94	871.80	1814.22	2168.24	18.0%	37.4%	44.7%
34490	43	89Zr-hCD133+	IB-hand	10	1974.18	272.53	18.41	293.04	150.35	9.41	303.32	22.29	2154.72	5017.28	1414.79	48 028.71	9.2%	2.6%	88.2%
33185	47	89Zr-hCD133+	IB-hand	10	98.32	16.65	9.08	15.13	5.32	3.35	10.8	0.09	3774.47	151.18	17.82	339.70	29.7%	3.5%	66.8%
31717	51	89Zr-hCD133+	IB-hand	10	1114.79	231.79	8.76	223.35	113.69	27.04	24.62	1.95	2862.3	2030.48	3074.18	5581.49	19.0%	28.8%	52.2%
33186	44	89Zr-hCD133+	IB-slow	10	276.65	65.92	4.13	18.21	6.53	2.16	0	0	3364.94	272.25	14.10	0	95.1%	4.9%	0.0%
33189	52	89Zr-hCD133+	IB-slow	10	1685.55	244.9	6.3	47.82	15.57	1.62	0	0	2949.48	1542.87	25.22	0	98.4%	1.6%	0.0%
33190	50	89Zr-hCD133+	IB-slow	10	1210.75	230.51	6.89	275.87	34.17	9.98	0	0	2466.73	1588.21	341.02	0	82.3%	17.7%	0.0%
33296	40	89Zr-hCD133+	IB-slow	10	986.39	168.28	4.93	13.83	3.86	1.6	0	0	2245	829.62	6.18	0	99.3%	0.7%	0.0%
32261	48.6	89Zr-hCD133+	IB-slow	10	2580.88	958.73	2.54	75.42	25.57	1.45	0	0	2120.79	2435.17	37.08	0	98.5%	1.5%	0.0%
33300	38	89Zr-hCD133+	IB-slow	10	1145.85	200.71	5.11	40.77	13.41	3.31	0	0	2087.17	1025.63	44.39	0	95.9%	4.1%	0.0%

Injection of ⁸⁹Zr-hCD133+ cells IV and IB

In contrast to ⁸⁹Zr-Df administration without cells, nearly all of the radioactivity (99.7%ID) was retained in the lungs following ⁸⁹Zr-hCD133+ cell IV-infusions, consistent with cell trapping within the alveolar capillaries (Figure 4B and Table 2).

After selective internal iliac artery ⁸⁹Zr-hCD133+ cell infusion (n=1), radioactivity remained in the iliac-BM for up to 1 h; thereafter radioactivity was observed to advance in the direction of arterial flow beyond the iliac-BM (Supplementary Figure S3) and did not remain completely confined to the marrow.

One hour following ⁸⁹Zr-hCD133+ cell hand IB-injection into two sites on the ipsilateral iliac bone, a significant proportion of radioactivity was observed in the lungs (44.7%ID) and locally in the surrounding muscle and soft tissue outside the iliac-BM, suggesting leakage into the venous circulation with lung trapping as well as cellular extravasation from the iliac-BM (37.4%ID) (Figure 4C).

Following ⁸⁹Zr-hCD133+ cell hand IB-injection via a single IB site, all animals (n=3) demonstrated some retention of radioactivity in the iliac-BM, although radioactivity was readily detected within the lungs (52.2–88.2%ID) consistent with rapid leakage into the circulation with cellular-retention in the lungs (Table 2 and Figure 4D). As similarly observed with high-rate IB-injection of contrast, hand-injection of ⁸⁹Zr-hCD133+ cells generated extremely high intra-marrow pressures that were substantially higher than those observed with rate-controlled slow IB-infusion (0.2 mL/min) (Figure 5).

Figure 5:

Systemic arterial and intra-marrow pressures over time during (A) hand IB-injection of ⁸⁹Zr-hCD133+ into the iliac-crest through a single site and (B) slow controlled non-manual infusion at 0.2 mL/min of ⁸⁹Zr-hCD133+ into the iliac-crest through a single site.

Remarkably, when the rate of ⁸⁹Zr-hCD133+ cell IB-injection was decreased to 0.2 mL/min at a single IB site, radioactivity at 1 h remained completely confined to the iliac-BM (82.3–99.3%ID) without any lung radioactivity being observed (n=6) (Figure 4E and Table 2), consistent with injected-cells being retained within the marrow cavity.

Discussion

Allogeneic HCT cures many malignant and non-malignant hematological conditions. Traditionally, allogeneic HPCs are IV-infused into the recipient. For donor-engraftment to occur, HPCs must migrate or home to the BM niche utilizing cytokine gradients and their endothelial cell adhesion molecules to attach to BM endothelium, with subsequent transmigration into the extracellular space where survival and proliferation signals are received (5). Although this process leads to donor-engraftment in the majority of transplants, HPCs must overcome considerable hurdles to survive and eventually engraft including the pulmonary first-pass effect, whereby the majority of administered HPCs are initially trapped in the pulmonary vasculature (6). While not a major impediment to allogeneic BM or mobilized blood stem-cell transplants where large numbers of HPCs are infused, this maybe indeed impede engraftment when limited numbers of HPCs are transplanted, such as when umbilical cord blood (UCB) transplants are utilized. In this setting, the minimal threshold CD34+ UCB cell-dose for successful engraftment is 1.5–1.7×10⁵ cells/kg recipient weight (7), which is more than one log less than the average transplanted dose of 5×10⁶ CD34+ cell/kg in patients receiving mobilized allografts. This lower cell-dose is the primary factor contributing to delayed donor-engraftment and the high incidence of graft-failure associated with UCB transplantation, both which increase mortality and limit the wider applicability of this transplant approach. This limitation has led investigators to employ the use of double UCB units in adult recipients of HCT (8,9) as well as explore methods to expand ex vivo cord blood HPCs prior to their transplantation (10,11). Furthermore, UCB HPCs have recently been shown to have an inherently decreased homing efficiency compared to their adult counterparts, owing to a variety of factors including, but not limited to, reduced fucosylation of surface CD44 (12). While ex vivo fucosylation of UCB may improve engraftment (13), utilizing a mechanical approach to retain HPCs in the marrow by direct IB delivery overcomes the need for ex vivo incubation and manipulation of thawed UCB HPCs.

An alternative strategy to improve the success of HCT procedures that utilize low numbers of transplanted HPCs involves bypassing the pulmonary first-pass attrition effect by directly injecting donor HPCs into the recipient BM. Although the application of IB-HCT has thus far been limited to only a few investigational UCB transplant trials in humans, the current prevailing technique for IB-injection involves propofol anesthesia followed by “hand-push” injection of 5mL aliquots through a standard 14 gauge Jamshidi needle into two separate sites on the iliac-crest, repeated bilaterally for a total infused volume of 20mL (14). Our results clearly show that multiple punctures in the iliac bone lead to cellular leakage into pelvic soft tissues upon the second ipsilateral injection, potentially negating the benefits of using a direct IB-injection approach to facilitate engraftment. Despite the use of a less than optimal IB-HCT approach, clinically successful engraftment has nevertheless already been reported in humans. More than half a century ago, the first reported case of successful allogeneic BM transplantation to cure BM failure employed directinjection of donor BM into the recipient’s sternum (15). More recently, Frassoni et al (14) and subsequently Okada et al (16) demonstrated successful engraftment in adult recipients of a single IB-transplanted UCB. However, a clinically meaningful improvement in the time to donor neutrophil engraftment compared to UCB HCT using a conventional IV approach was not clearly established by either investigator. Other attempts at IB-HCT employing larger volumes of transplanted cells did not demonstrate any advantages compared to a conventional IV approach (17,18). Hagglund et al reported no difference in outcomes between the IV- and IB-route in allogeneic BM transplantation. In their analysis, ^99mTc labeled BM cells IBinfused in a volume of approximately 1000mL resulted in radioactivity identified immediately in the heart following injection (18). Brunstein et al (17) also reported no advantage for UCB-HCT using the IB-route in terms of achieving unit dominance compared to the IV-route in the setting of a double-UCB transplant.

In contrast to small-animal studies, these pilot trials in humans failed to demonstrate superior engraftment utilizing the IB-route of HCT (19-21). Xenotransplantation of human UCB HPCs into mice utilizing direct IB-injection achieves superior long-term engraftment and the ability for secondary SCID mouse repopulation (22-24). These preclinical small-animal models cannot predict engraftment kinetics in humans undergoing IB-HCT. BM blood-flow, cardiovascular physiology and anatomical size are not comparable between mouse and man. In humans, the IB-route of crystalloid infusion can rapidly restore circulating volume and blood pressure in the setting of hypovolemic shock. In fact, in large-mammal models, it has been shown that IB-infusion of crystalloid can achieve peripheral-to-central circulation times comparable to those achieved by the IV-route (25,26). Additionally, there is a substantial risk of fat-embolism following rapid IB-infusions (27,28).

We selected the porcine model used in this study because of its comparable marrow blood flow, cardiovascular physiology and pelvic anatomical size to humans (29). Using radiolabeled HPCs, we tracked injected-cells in vivo and optimized conditions for retaining HPCs in the BM using the IB-route. These experiments required purified HPCs to demonstrate proof of principle that the mechanical refinement of this IB infusion technique will result in retention of human HPCs in marrow. We chose to utilize purified human HPCs which were available in large numbers for these experiments, given the homology between receptor ligands expressed on human and porcine endothelial cells for their cognate receptors expressed on human HPCs (30) that allows human HPCs to traffic across porcine endothelium into swine bone marrow (31). Although investigators have shown that 5–15% of labels used for cells can be incorporated into host macrophages (32), macrophage engulfment in the first hour likely did not confound these experiments as it would have been below the threshold for radioactivity detection.

Here we demonstrate a purely mechanical method to improve the retention of HPCs in the marrow following an optimized IB delivery technique. Furthermore, we developed and optimized a novel cell-labeling technique using ⁸⁹Zr and protamine that is readily applicable for clinical trials that incorporate cell tracking of transplanted human HPCs. Using ⁸⁹Zr cell-labeling, we compared cellular-retention in the BM with different IB-injection techniques utilizing a PET/CT scanner. IV-injection of ⁸⁹Zr-Df alone revealed rapid renal excretion whereas with IV-injection of ⁸⁹Zr-labeled HPCs revealed extensive early lung trapping. To maximize IB-retention, we optimized the injection needle to deliver cells at right angles to the long axis using side fenestrations. By measuring injection-pressures continuously, by limiting injected-volumes and flow-rates, and by monitoring the distribution of injected-cells, we were able to determine optimal injection parameters that resulted in maximum retention of HPCs in the iliac-BM. We demonstrated that high intra-marrow pressures led to extravasation into surrounding soft tissues outside the iliac-BM and rapid venous drainage from the marrow, both sequelae that are undesirable for IB cell retention. High intra-marrow pressures also elevated PASP, likely as a consequence of procedure-associated pulmonary embolisms. While this complication has not yet been reported in current human clinical trials involving IB HCT, it does not exclude the possibility that subclinical pulmonary embolisms may have occurred. We also found that when two insertions were made into the ipsilateral iliac crest, the second injection often leaked through the first injection-site regardless of pressure or distance between the sites. Thus, we demonstrate that the maximum proportion of injected-HPCs retained in the marrow can be achieved when injection-pressures are controlled using a pump that limits the flow-rate to ≤0.2mL/min and the injected-volume to 10 mL into a single ipsilateral-site in the iliac crest. In clinical application, this will allow for 20 mL of cellular product to be split into two aliquots for infusion into both iliac crests, each through a single site. The infusion time on either side will be less than an hour, well within the time period before noticeable loss of CD34+ cell viability is observed with washed or unwashed products (33). This technique offsets limitations related to hand cell-injection where high-flow rates from the syringe into the marrow result in turbulent flow, dissipation of energy and shear-stress on the HPC cell-membrane, all which contribute to cell disruption and viability loss (34).

The successful ⁸⁹Zr-labeling of HPCs was pivotal in this study for monitoring their fate following injection. This method utilized a very simple mixture of ⁸⁹Zr and protamine with cells. As cellular uptake of the ⁸⁹Zr-protamine complex was self-limited, only miniscule doses of radioactivity (approximately 10 μCi per 2×10⁸ hCD133+ cells) was retained by the cells, preserving a high viability among labeled-cells with little chance of cell-death occurring from radiation effects. In fact, the viability of the labeled-cells at 14 h was comparable to unlabeled-cells in a single experiment. Our unpublished data of long-term effects of labeling with ⁸⁹Zr in other mononuclear cell populations suggests that the phenotype and function of cells are unaffected by the labeling process. The ability to accurately detect and image with such a low dose of radioactivity greatly reduces the risk of time-dependent cytotoxicity seen with the larger doses of ¹¹¹In-oxine required for cell-labeling (1). The sensitivity of PET/CT scanners enables detection of this small dose of cellular radioactivity. ⁸⁹Zr has several advantages for this task: it has a sufficiently long half-life (78.4 h) to track cells for up to a week and its physical properties produce relatively high spatial resolution (35) compared to other positron emitters. Because PET cameras are highly sensitive and have relatively better resolution than SPECT cameras and because there is no background signal in the recipient, even very tiny amounts of radioactivity can be detected. Moreover, since PET images are superimposed on high-resolution CT images, it is possible to accurately localize the injected-cells with a high degree of certainty.

This study was not designed to demonstrate that confining transferred HPCs to the BM results in improved engraftment following HCT. However, our data does support a methodology of HPC labeling that can be used to optimize, standardize and validate IB-injection in humans, removing uncertainty and inconsistency in cellular-delivery methods for patients undergoing this investigational transplant approach. Human clinical trials are required to demonstrate whether this optimized IB-transplantation method reported results in superior donor-engraftment in humans undergoing UCB HCT. In a syngeneic mouse model, engraftment and hematopoietic reconstitution levels were very similar between IB- and IV-transplanted mice (36). However, in congenic HPC transplantation in rats although IB-injection resulted in rapid delivery to the bloodstream, there still was reduced pulmonary sequestration compared with IV-injected HPCs, suggesting that first-contact with the BM resulted in improved recruitment back to their hematopoietic niche (37).

There may be other advantages to utilizing an IB-approach to transplantation. In the largest human clinical trial of UCB IB-HCT in acute leukemia, none of the 26 evaluable patients developed grade III–IV acute-GVHD. This is in contrast to the 9% risk of severe acute-GVHD reported in other UCB HCT clinical trials (38,39). Although antithymocyte globulin was used in the preparative regimen and could have contributed to the lower incidence of acute-GVHD compared to conventional UCB HCT, this immunological advantage was also demonstrated in small-mammal models of IB-HCT. In a mismatched murine model of allogeneic HCT where C57BL/6 recipient mice received BALB/c donor BM and lymphocytes either IB or IV, the IB route was associated with an abrogation of GVHD (40). Similarly, in a rabbit model of haploidentical HCT, engraftment was improved and GVHD reduced in IB compared to IV transplant recipients (41). In another study, Brown Norway rats were the recipients of hind limbs transplanted from donor Fischer 344 rats together with an IB transplant of donor bone marrow cells after fludarabine and low dose irradiation conditioning. Recipients achieved complete donor lympho-hematopoietic engraftment and transplanted limbs were accepted for more than 1 year without any clinical signs of rejection. Lymphocytes harvested from transplanted recipient rats showed tolerance to both donortype and recipient-type major histocompatibility complex determinants in mixed lymphocyte reaction assays but maintained a preserved response against third party (42). Thus, optimized IB-HCT in humans could offer multiple immunological advantages in addition to the potential of improving engraftment in the context of transplants that utilize small number of donor HPCs or for induction of tolerance in recipients of combined bone marrow solid organ transplants.

In summary, we show that conventional methods utilized for IB-HCT in humans are ineffective in maintaining cells in the marrow space. Utilizing a novel cell-labeling technique, we developed a method to optimize the retention of HPCs in the BM, which was contingent upon real-time intra-marrow pressure monitoring and reducing the infusion rate and volume. A key component to the success of this technique was the use of a modified needle with side fenestrations allowing for real-time continuous monitoring of intra-marrow pressures during IB-injection, as maintaining low IB-pressures was found to be essential to retain cells within the marrow. The development of a device that uses continuous pressure feedback monitoring to regulate the infusion rate to maintain low intra-marrow pressure is key to this work and will need further testing to determine if the maximum volume of cells that can be delivered varies depending on patient specific variables such as pelvic size and/or the specific location where IB delivery of cells occurs. The injection methods used in this study have establish an optimized technique that is currently being used to investigate the impact of IB-HCT on engraftment in a non-human primate model, and further lays the foundation for future clinical trials that systematically explore IB-HCT in humans.

Supplementary Material

Supple figs

Figure S1: CT angiography of the left hemipelvic terminal branches of the abdominal aorta in Susscrofa. The right external and internal iliac arteries were reversibly occluded by external vessel occluders (arrows) during contrast injection.

Figure S2: 3D reconstruction following second hand IB-injection of contrast with two access sites on the right iliac-crest. Leakage is observed from the first injection-site with run-off (arrow).

Figure S3: Fusion PET/CT coronal images following right iliac artery intra-arterial infusion of⁸⁹Zr-hCD133+ at (A) 1 h, (B) 2 h, and (C) 3 h post infusion. Radioactivity is noted to advance from within the iliac-BM into the surrounding muscle in the direction of arterial blood flow.

SuppVideo1

Video S1: Dynamic 3D CT angiography during contrast IB-injection into the pelvis at 1 mL/s demonstrating leakage at the injection-site and rapid drainage into the central venous system via the iliac vein.

SuppVideo2

Video S2: Dynamic 3D CT angiography during second hand IB-injection of contrast with two access sites on the right iliac-crest. Leakage is observed from the first access site during injection into the second access site.

Abbreviations:

%ID

percent-injected dose

3D

3-dimensional

ACT

activated clotting time

BM

bone marrow

CD

cluster differentiation

Ci

Curie

CT

computed tomography

Df

deferoxamine

Fr

French

G-CSF

granulocyte-colony stimulating factor

GVHD

graft-versus-host disease

hCD133+

human CD133 positive

HCT

hematopoietic cell transplantation

HPC

hematopoietic progenitor cell

IB

intrabone

In

indium

IV

intravenous

KCI

potassium chloride

M

molar

MSC

mesenchymal stromal cell

PASP

pulmonary artery systolic pressure

PBS

phosphate buffered saline

PET

positron emission tomography

SCID

severe combined immunodeficiency

SPECT

single-photon emission computed tomography

Tc

rechnetium sestamibi

UCB

umbilical cord blood

VOI

volume of interest

Y

Yttrium

Zr

Zirconium