Cell size and velocity of injection are major determinants of the safety of intracarotid stem cell transplantation

Abstract

Intracarotid transplantation has shown potential for efficient stem cell delivery to the brain. However, reported complications, such as compromised cerebral blood flow (CBF), prompted us to perform further safety studies. Glial-restricted precursors (GRPs) and mesenchymal stem cells (MSCs) were transplanted into the internal carotid artery of rats (n=99), using a microcatheter. Magnetic resonance imaging was used to detect post-transplantation complications, including the development of stroke, for the following experimental variables: cell size, cell dose, cell infusion velocity, delay between artery occlusion and cell infusion, discordant versus concordant xenografting, and intracarotid transplantation with preserved versus compromised blood flow. Immunocompatibility and delayed infusion did not affect the number of complications. An infusion velocity over ⩾1 mL/minute often resulted in stroke (27 out of 44 animals), even with an infusion of vehicle, whereas a lower velocity (0.2 mL/minute) was safe for the infusion of both vehicle and smaller cells (GRPs, diameter=15 μm). Infusion of larger cells (MSCs, diameter=25 μm) resulted in a profound decrease (75±17%) in CBF. Stroke lesions occurred frequently (12 out of 15 animals) when injecting 2 × 10⁶ MSCs, but not after lowering the dose to 1 × 10⁶ cells. The present results show that cell size and infusion velocity are critical factors in developing safe protocols for intracarotid stem cell transplantation.

Introduction

The intravascular route of stem cell delivery has met with increasing interest because of the minimally invasive nature of the procedure and the potential for broad cell distribution. Recent reports^{1, 2} have revealed positive effects of intravascular cell transplantation in animal models of neurologic disorders. The evaluation of neural cell distribution after intravenous cell transplantation has shown that most of the cells are initially entrapped within the lungs and do not travel to the brain.³ It has been hypothesized that an intraarterial approach would be a more efficient route of cell delivery to the brain, as this approach avoids the pulmonary circulation. This approach is particularly attractive when selecting and sorting cells for adhesion molecules, which can enhance cell homing and therapeutic outcome.⁴ Moreover, it has been recently shown that transfection of adhesion molecules in progenitor cells results in a dramatic increase of their homing to inflamed brain endothelium.⁵

Direct injection of cells into arteries supplying blood to the brain naturally raises safety concerns. Adverse events, including high mortality, have been reported in some,^{6, 7} but not all,^{8, 9} animal studies. One study indicated that microembolisms may occur when cells are injected using a microcatheter technique, and complications can be eliminated by the use of a micro needle.¹⁰ However, even that technique is not devoid of disadvantages, including cell precipitation in the syringe/needle, complicating cell dosing or bleeding after removal of the needle from the artery.

Obviously, there are still many unresolved issues that compromise the efficacy and safety of intraarterial cell injection; thus, careful optimization of that procedure is needed. Noninvasive imaging of both injected cells and monitoring the status of brain tissue is highly desirable for optimization of arterial cell infusion and cerebral homing. One such technique is magnetic resonance imaging (MRI).^{5, 6} A catheter-based intraarterial injection technique is ideally suited for that purpose. We therefore explored, in a systematic fashion, the factors that determine the safety of intracarotid cell delivery using a microcatheter technique.

Our previous studies with intraarterial transplantation of relatively large size rat MSCs⁶ resulted in frequent complications attributed to embolisms, which prompted us to study this intravascular delivery with smaller (leukocyte size) glial-restricted precursors (GRPs), with the goal of minimizing the risk of microembolic stroke. Once a method for the safe infusion of GRPs was established, we investigated whether the procedure is safe for widely used and commercially available, but large-sized, human MSCs (hMSCs).

Materials and methods

Cells for Transplantation

Primary GRPs were derived from midgestation mice¹¹ or human fetuses (Q Therapeutics, Salt Lake City, UT, USA). Cells were immortalized using lentivirus encoding the SV40 large T-antigen, and were selected with puromycin. They were maintained in serum-free Dulbeccos's modified Eagle's-F12 medium supplemented with N2, B27, bovine serum albumin, and basic fibroblast growth factor. Human mesenchymal stem cells (MSCs; PT-2501, Lonza, Basel, Switzerland) were maintained in PT-3001 growth medium (Lonza).

Before transplantation, cells were trypsinized, spun down, resuspended in phosphate-buffered saline (PBS), and filtered through a 40-μm filter (catalog number 352340, BD Falcon, San Jose, CA, USA) to obtain a single-cell suspension. The concentration was adjusted to 1–8 × 10⁶ cells/mL. For experiments that evaluated the detection of transplanted cells with an MRI (initial experiment; n=15), cells were labeled overnight with 25 μg/mL Molday ION Rhodamine B (CL-50Q02-6A-50, BioPAL), a superparamagnetic iron oxide formulation.

For the measurement of cell size, cells were trypsinized, pelleted, resuspended in media, and pipetted in a hemocytometer. Images were then acquired at × 20 magnification (Olympus BX40, Center Valley, PA, USA). The cell diameter and circularity were measured with Photoshop (Adobe), and the cell area and volume were calculated (Excel, MS) for each individual cell type (n=60 cells each). The experiment was performed twice.

Intracarotid Cell Transplantation

All experimental procedures were in accordance with the guidance provided in the Rodent Survival Surgery manual, and were approved by our Institutional Animal Care and Use Committee. A total of 99 male Sprague-Dawley rats weighing 250 g were used throughout all the experiments, with different experimental conditions (see Table 1). Using inbred naive animals obviated the need for their randomization. Animals were anesthetized with 2% isoflurane and stabilized on a surgical table in the supine position. After a paramedian skin incision, the common carotid artery (CCA), the external carotid artery (ECA), and the internal carotid artery (ICA) were exposed within the carotid muscular triangle. The occipital artery branching off the proximal segment of the ECA was dissected and coagulated, and the pterygopalatine artery branching off the ICA was ligated. Further procedures differed depending on the selection of CCA or ECA for cannulation. For the CCA route, the proximal segments of the ECA and CCA were ligated, and a vascular clip (FT 180T, Aesculap, Center Valley, PA, USA) was applied to the ICA proximal to the pterygopalatine artery. Incision into the CCA was performed using microscissors just distal to the ligature, the catheter was inserted into the CCA, and the suture was tightened on the artery over the catheter to prevent blood outflow. The clip was removed just before cell injection. For the ECA route, vascular clips were applied to the CCA and ICA, and the ECA was ligated at the maximum distal position. After the incision of the ECA in close proximity to the ligature, the catheter was inserted into the stump of the ECA without occluding the ICA, assuring proper blood flow through the CCA and the ICA. The suture was tightened on the ECA over the catheter to prevent blood outflow. Both clips were removed immediately before cell injection.

Table 1. Occurrence of stroke under different experimental conditions.

Cell type or vehicle	CCA closure one day	Velocity of injection	Route	Cell dose	Stroke (n of animals)	No stroke (n of animals)
hGRPs—SPIO-labeled	No	3.0 mL/minute	CCA	2 × 106	11	4
mGRPs					6	2
	Yes				10	5
PBS	No			No cells	4	2
		1.0 mL/minute			2	4
		0.2 mL/minute			0	6
mGRPs				2 × 106	0	6
hGRPs					0	6
MSCs					7	1
			ECA		5	2
				1 × 106	0	6
			CCA		0	6
mGRPs				8 × 106	0	3
Total n of animals					45	53

Abbreviations: CCA, common carotid artery; ECA, external carotid artery; hGRP, human glial-restricted precursor; mGRP, mouse glial-restricted precursor; MSC, mesenchymal stem cell; PBS, phosphate-buffered saline; SPIO, superparamagnetic iron oxide.

After cannulation of the appropriate artery, 1 to 8 × 10⁶ cells in 1 mL of 10 mmol/L PBS, pH 7.4, were delivered using an infusion pump (QSI, Stoelting, Wood Dale, IL, USA). PBS vehicle (1 mL) without cells was injected as a control. This volume was delivered over three time spans (injection speeds) of 20 s, 1 min, and 5 min. After transplantation, the catheter was removed, the arteriotomy site was ligated to prevent bleeding, the skin was closed, and the animal was allowed to recover from anesthesia. In one of the animals with CCA cannulation, the arteriotomy site was sutured and blood flow was restored to the CCA.

Monitoring of Cerebral Blood Flow with Laser Doppler

For laser Doppler flowmetry, the transplantation procedure was preceded by implantation of a fiber optics probe to the skull. The skin was incised in the temporal area, the temporal muscle was dissected, and the tip of the optic fiber was glued (Loctite 4161, Henkel Corporation, Westlake, OH, USA) to the temporal bone. The laser Doppler signal was recorded and analyzed using moorVMS-PC V3.1 software (Wilmington, DE, USA). After cell transplantation, the optic fiber was detached from the temporal bone, and after hemostasis the wound was sutured and closed.

Laser Doppler flowmetry provides a relative measurement of the cerebral blood flow (CBF) and, therefore, values are reported about baseline. The mean value of CBF during the 5 minute before cell infusion (baseline), the mean CBF value during the 5 minute of the cell infusion, and the time to restore CBF to the pretransplantation level were measured. The decrease in CBF signal during cell transplantation was expressed as a percent of baseline.

Bioluminescent Imaging

Bioluminescent imaging was performed immediately and 24 hours after cell infusion using an IVIS Spectrum optical imaging device (Caliper LifeSciences, Waltham, MA, USA). Luciferin was administered intraperitoneally at 150 mg/kg and bioluminescent imaging was performed every 5 minutes for up to 30 minutes to reach the signal peak. The exposure time was set at 1 minute, with the data represented as photon flux (photons/s).

Magnetic Resonance Imaging

MRI test was performed using a 9.4T or our upgraded 11.7T scanner (Bruker Biospin, Billerica, MA, USA). T2*-weighted scans were used to detect magnetically labeled cells, and a T2-weighted (T2-w) multislice multiecho sequence was used for stroke evaluation. For imaging at 9.4T, isoflurane-anesthetized animals were imaged using a custom-built 35 mm volume coil with the following parameters: for T2* fast low-angle shot sequence—repetition time=300 ms, echo time (TE)=7 ms, flip angle=45°, acquisition time=10 minutes 14 seconds; and for the multislice multiecho sequence—repetition time=2,000 ms, TE=12 to 60 ms flip angle=180°., acquisition time=8 minutes 32 seconds. T2w-images with TE=12 ms were used for anatomic evaluation, whereas scans with a longer TE of 60 ms were used for monitoring of stroke development.

For imaging at 11.7T, isoflurane animals were scanned using a Bruker 15 mm planar surface coil and the following parameters: field of view=1.35/2.17, T2-w RARE sequence (repetition time=5,000 ms, TE=30 ms, flip angle=180°, RARE=8, acquisition time=2 minutes 40 seconds). This T2-w sequence was used for both anatomic evaluation and stroke detection. An MRI specialist, masked to experimental conditions, confirmed the occurrence of stroke. The animals were killed at the 1-day follow-up imaging session.

Statistical Analysis

Continuity-adjusted χ²-statistics (PROC FREQ, SAS) were used for categorical data (occurrence of stroke). A restricted maximum likelihood approach (PROC MIXED, SAS, Cary, NC, USA) was employed for calculations of continuous data (cell size and CBF by laser Doppler).

Results

Nonoptimized Intracarotid Cell Injections Cause Microembolisms and Lacunar Strokes

In the initial experimental transplantation paradigm, the ECA was ligated together with the pterygopalatine artery and the proximal segment of the CCA. Superparamagnetic iron oxide-labeled human GRP (hGRP) cells (2 × 10⁶) suspended in 1.0 mL of PBS were then injected into the CCA via a microcatheter (Cole-Parmer, Vernon Hills, IL, USA, PTFE number 30) at a rate of 3 mL/minute, which corresponds to the blood flow velocity in the CCA.¹² Bioluminescent imaging (Figure 1A) and MRI (Figure 1B), performed directly after cell transplantation, confirmed that the infused cells reached the brain. As expected, T2-w MRI did not detect stroke lesions at this early time point (Figures 1C and 1D). Follow-up bioluminescent imaging (Figure 1E) and MRI images (Figure 1F) performed 24 hours later showed a substantial loss of hypointense signal, indicating clearance of superparamagnetic iron oxide-labeled cells from the brain. T2-w images with a short TE=12 ms provided good anatomic detail, and did not show significant abnormalities (Figure 1G). Longer echo images with TE=60 ms were used to identify stroke lesions (Figure 1H), which occurred in 11 of 15 animals (Table 1). Most of these strokes were lacunar and were located within or in close proximity to the corpus callosum (Figure 1H), whereas the cortex and the deep structures were largely spared. Because of the high complication rate with the above procedure it was considered unsafe, and further experiments were designed to identify the underlying basis for this frequent stroke development.

Figure 1.

Bioluminescent imaging (BLI) and magnetic resonance imaging (MRI) (9.4T) after transplantation of 2 × 10⁶ luciferase- and superparamagnetic iron oxide (SPIO)-labeled human glial-restricted precursors (hGRPs) in 1 mL of phosphate-buffered saline (PBS) at a velocity of 3 mL/minute. Directly after transplantation (A–D), the cell signal is visible on BLI (A) and MRI (B). A normal morphology without edema (C) and the occurrence of stroke (D) is observed immediately after transplantation. MRI at 1 day after injection (E–H) shows a dramatic decrease in cell signal in the brain for both BLI (E) and MRI (F). Despite an unaltered brain morphology (G), lacunar strokes are visible on late echo T2-weighted (T2-w) MRI (H, white arrows).

Effect of Immunologic Discordance

The first factor we investigated was the probability of brain artery thrombosis induced by the high immunologic discordance between infused hGRPs and the rat host, and the lesser mismatch for the mouse GRPs (mGRPs). Using identical experimental conditions (dose, route, velocity of injection, etc.), there was no difference between the two xenograft transplants, with a high incidence of 11/15 (73%) and 6/8 (75%) animals for hGRPs and mGRPs, respectively (P=1.0). Hence, the hypothesis of a major role for immune system-related thrombosis in stroke formation was rejected.

Effect of Procedure-Dependent Changes in Cerebral Blood Flow

We further hypothesized that the ligation of the proximal part of the CCA performed immediately before transplantation could result in a transient change in the CBF, caused by a sudden rerouting of cerebral blood circulation from other cerebral arteries. Such a change in CBF could cause a sluggish flow, leading to insufficient oxygenation of the brain tissue or thrombosis, both potentially resulting in the development of a stroke. Thus, we performed the ligature of the CCA, the ECA, and the pterygopalatine artery 1 day before mGRP cell injection, to allow time for a new pattern of circulation to be established in the affected hemisphere. MRI just before cell injection demonstrated that ligation itself does not lead to the formation of stroke lesions. Unfortunately, similar stroke lesions were observed 24 hours after mGRP injection in 10/15 (67%) animals despite a modification of the procedure. Thus, the ligation of arteries 1 day in advance did not contribute to a decrease in the complication rate (P=1.0).

Effect of Restoration of Blood Flow Following Transplantation

Reasoning that the lack of blood flow in the CCA after cell transplantation could potentially affect CBF parameters, we restored the blood flow after cell transplantation in one of the experimental animals. This was accomplished by performing the technically challenging procedure of placing a suture in the arterial wall at the site of arteriotomy, reestablishing the patency of the CCA. However, despite restoration of blood flow after transplantation, we still observed the occurrence of lacunar strokes and, thus, this procedure was discontinued.

Velocity of Injection

As the above-mentioned strategies did not prevent complications, we next investigated the effect of vehicle (PBS) injection without cells. Laser Doppler flowmetry was used to measure CBF. Injection of PBS at a velocity of 3 mL/minute resulted in a substantial drop in CBF throughout the injection period. Notably, MRI at 24 hours follow-up revealed lacunar strokes in 4/6 (67%) animals with a distribution similar to the previous cell injection experiments (Figures 2A–2C). Using the same experimental conditions, the lesion occurrence rate in this vehicle control group was similar to that of the cell transplantation group (P=1.0). This data thus indicate that not the cells but the infusion procedure itself is a major determinant for stroke development. Hypothesizing that the velocity of fluid injection might be a major contribution, we adjusted the rate downward to 1 mL/minute. Under these conditions, the drop in CBF on laser Doppler was much smaller, but 24-hour follow-up MRI images still showed lacunar strokes in 2/6(33%) animals. We further decreased the infusion velocity to 1.0 mL over 5 minutes (i.e., an infusion rate of 0.2 mL/minute), which did not result in a decrease in CBF as measured with laser Doppler flowmetry. MRI testing at 24-hour follow-up did not detect any strokes (0/6 animals). Eliminating the occurrence of stroke attributable to the infusion procedure itself prompted us to revisit the experimental conditions of infusing mGRPs at a dose of 2 × 10⁶ cells. Indeed, cell infusion at a rate of 0.2 mL/minute did not result in a reduction in CBF on laser Doppler (Figure 3A), and no strokes were detected on MRI after 24 hours (0/6 animals). The adjustment of the injection velocity completely eliminated the complications, and this improvement was statistically significant compared with mGRPs transplanted at a rate of 3 mL/minute (P=0.024). We then tested these experimental conditions for hGRP transplantation and also found no decrease in CBF, with no stroke occurrence in any of the six animals, and these results were also significantly different compared with an infusion rate of 3 mL/minute (P=0.01). Thus, the velocity of cell infusion is a major determinant of the safety of intracarotid transplantation.

Figure 2.

Detection of lacunar strokes by T2-weighted (T2-w) magnetic resonance imaging (MRI) (11.7T). Lacunar strokes (white arrows) in representative rats after injection of 1 mL of phosphate-buffered saline (PBS) at a rate of 3 mL/minute (A–C) and mesenchymal stem cells (MSCs; 2 × 10⁶ in 1 mL at a velocity of 0.2 mL/minute (D–F)). Note the similar pattern of stroke locations.

Figure 3.

Laser Doppler cerebral blood flow (CBF) measurements. CBF during injection of glial-restricted precursors (GRPs) remained unchanged (A), whereas there was a sudden and dramatic drop in CBF during the injection of human mesenchymal stem cells (hMSCs), with a subsequent recovery of CBF to baseline levels (B). The start of cell injection is indicated on the x axis as t=0, with the duration of the infusion period (5 minute) indicated by arrows. PU, perfusion units that are a relative measure of CBF.

Cell Size

GRPs with a diameter of approximately 15 μm are relatively small; thus, we extended our analysis with hMSCs that are widely used for transplantation studies. MSCs have, on average, a diameter of 25 μm and are thus one of the largest cells in the body. From the cellular diameter and circularity, the surface area can be calculated to be 174±24, 234±30, and 465±20 μm² for mGRPs, hGRPs, and hMSCs, respectively (Figure 4). These differences were statistically significant (F=161.79, P<0.0001). The cellular volume calculations were 7,344 μm³ for hMSCs versus 1,695 μm³ volume for mGRPs, i.e., a 4.33-fold difference.

Figure 4.

Measured cell size (area in μm²) for the three cell types (n=60 each) used in this study. Values are expressed as mean±s.d. The experiment was performed twice.

Transplantation of 2 × 10⁶ hMSCs induced a dramatic decrease in CBF that lasted throughout the 5 minutes of infusion, and then recovered to the baseline level (Figure 3B). One-day follow-up MRI images showed lacunar strokes in seven of eight animals (88%), with the lesions being distributed primarily in proximity to the white matter structures (Figures 2D–2F), similar to those observed with fast-velocity saline injections (Figures 2A–2C). The difference between the transplantation of hGRPs and hMSCs for both stroke rate (P=0.007) and decrease in CBF was highly significant. The CBF decrease was very high for hMSCs (84.05±3.07%), whereas for hGRPs it was negligible (2.61±7.93% P=0.0001). Thus, cell size is also a major determinant of the safety of intracarotid transplantation.

Effect of Different Routes of Cell Delivery

We hypothesized that the closure of the CCA might decrease the ipsilateral cerebral blood pressure (CBP) during the transplantation procedure. This could result in large MSCs becoming trapped by the cerebral circulation and causing a stroke, whereas small hGRPs are able to pass through the circulation. To avoid the potential decrease in CBP, we injected the cells through a catheter inserted into the stump of the ECA, thus preserving the flow in the CCA during and after cell transplantation. Despite this modification, stroke lesions occurred in five of seven animals (71%). Thus, the preservation of the flow in the CCA during intracarotid cell injection does not lower stroke incidence compared with the permanent closure of the CCA to the blood flow during and after transplantation (P=0.9).

Number of Cells Injected

As the complications of MSC intracarotid delivery proved to be largely related to their cell size, we further investigated whether a reduction of the total cell dose would improve the safety of the procedure. We reduced the number of injected MSCs from 2 × 10⁶ to 1 × 10⁶ cells and infused the cells through the ECA, preserving the patency of the CCA during and after cell injection. The 1-day follow-up MRI testing did not show an occurrence of stroke in any of the six animals. This improvement that was obtained by halving the cell dose was statistically significant (P=0.04). We then injected 1 × 10⁶ MSCs into the CCA, with the proximal CCA permanently ligated, thus eliminating blood flow in this artery. In this experiment, there were also no stroke lesions visible on the MRI in any of the six animals. Thus, the total number of injected cells, and not the patency of the CCA during intracarotid transplantation determines the safety of MSC transplantation (P=0.007).

Finally, to evaluate the relative contribution of the injected cell dose versus cell size on the safety of the procedure, we dramatically increased the dose of mGRPs to 8 × 10⁶ cells, with the total cell volume being comparable to that of 2 × 10⁶ MSCs. When transplanting this large dose of mGRPs, we did not observe a decrease in CBF nor a stroke occurrence in any of the animals. Thus, our data indicate that intracarotid delivery of small cells, such as mGRPs, is safe, even at large doses. Safety considerations should be incorporated, particularly when using larger cells, such as MSCs, and the dose of these cells needs to be carefully adjusted.

CBF Measurements

The injection of 1 mL hGRP solution at a rate of 0.2 mL/minute did not result in a reduction in CBF, as measured by laser Doppler flowmetry. The observed mean decrease in CBF by 2.61±7.93% was negligible and statistically nonsignificant (P=0.39). However, the infusion of the same volume of MSCs (mean for both doses and both routes) revealed a statistically significant decrease in CBF by 74.0±16.2% (P<0.0001). The extent in CBF decrease did not statistically differ between either route or dosage of MSCs (Table 2). The mean time for recovery of CBF to baseline levels after the infusion of MSCs was 5.6±3.6 minute. This was also not dependent upon delivery route or cell dose (Table 2).

Table 2. CBF values measured using by laser Doppler flowmetry.

Variable	Effect	Route	Dose	Mean	s.d.	F value	P value
Decrease in CBF (%)	Route	CCA		80.554	10.864	3.44	0.08
		ECA		68.571	18.957
	Dose		2 × 106	76.117	15.578	0.85	0.37
			1 × 106	71.919	18.096
	Route*dose	CCA	2 × 106	84.048	3.067	0.01	0.922
		CCA	1 × 106	78.226	13.803
		ECA	2 × 106	71.586	21.196
		ECA	1 × 106	64.351	21.196
Time to restore CBF (minutes)	Route	CCA		5.921	2.609	0.11	0.744
		ECA		5.321	4.441
	Dose		2 × 106	5.857	4.588	0.07	0.793
			1 × 106	5.379	2.641
	Route*dose	CCA	2 × 106	4.835	3.061	1.89	0.186
		CCA	1 × 106	6.645	2.249
		ECA	2 × 106	6.538	5.556
		ECA	1 × 106	3.860	2.413

Abbreviations: CBF, cerebral blood flow; CCA, common carotid artery; ECA, external carotid artery.

Discussion

Contemporary treatments for neurodegenerative diseases can be invasive and charged with possible complications, but the risks should not outweigh the benefits. Although intracarotid cell delivery is theoretically a very promising method of neurotransplantation, preclinical studies have raised concerns about its safety. Indeed, we initially observed an unacceptably high rate of stroke complications after this route of injection. These unwanted effects could potentially limit the use of the intracarotid route in preclinical investigations, or even lead to complete abandonment of this cell-delivery route for clinical use. This prompted us to study the risks of intracarotid cell transplantation and determine which factors are most relevant for developing safe injection protocols.

Microvascular thrombosis is a major complication of solid organ xenografts^{13, 14} and, thus, we first hypothesized that the presence of xenogeneic cells in cerebral vessels may be a factor contributing to the thrombosis and subsequent strokes that we observed in our hGRP grafts. As human-to-rat is a discordant xenotransplantation, it activates the immune system more profoundly than concordant xenografts.¹⁵ Here the activation of complement by an immune reaction can stimulate coagulation.¹⁶ To test this hypothesis, we performed phylogenetically closer, concordant mice-to-rat transplantations. We found that this paradigm did not decrease the complication rate, and rejected the hypothesis that the immune system had a leading role in the development of stroke after intracarotid cell transplantation.

Next, we investigated the influence of CCA ligation on stroke occurrence. Because of the extensive cerebrovascular reserve capacity, a unilateral occlusion of the carotid artery was found to be safe in normal rats. However, cerebral hypoperfusion ischemia is more profound in the hemisphere ipsilateral to the CCA closure, and the delay of ischemia induction neutralizes this effect, probably because of long-term vascular adaptations to improve the collateral circulation.¹⁷ We therefore investigated the effect of CCA ligation one day before cell transplantation, and found no effect on the rate of stroke complications.

We then investigated the relative effect of the velocity of cell injection using laser Doppler flowmetry for measuring CBF. We determined the infusion rate to be critical for the safety of the transplantation procedure. Although an infusion rate of 3 mL/minute resulted in stroke lesions, even with an injection of saline, a reduction of the infusion rate to 0.2 mL/minute entirely eliminated that complication. Furthermore, we observed that the infusion of large doses of small cells (GRPs) do not compromise the CBF, whereas larger cells (MSCs) cause a 75% decrease in CBF. The diameter of GRPs (13 to 15 μm) is comparable to that of endogenous circulating rat leukocytes,¹⁸ and rat capillary vessels can, therefore, easily adapt to this transplanted cell size, assuring continuous blood flow. In contrast, the rat microcirculation is not naturally adaptable to receiving MSCs with a much larger diameter of 25 μm, leading to entrapment of MSCs within the cerebral microcirculation. We observed this reduction in CBF with both rat MSCs in our previous study⁶ as well as in current experiments with hMSCs. Surprisingly, cell infusion via the ECA, which preserves blood flow in the CCA, did not decrease the stroke rate, and also did not alleviate the drop in CBF. These findings indicate that the circle of Willis is well-developed in rats, and that there is no clear advantage of preserving the patency of the CCA. The decrease in the dose of MSCs to 1 × 10⁶ cells completely eliminated the occurrence of stroke. Although the exact mechanism of the development of the characteristic pattern of lacunar strokes after excessive cell injection is unknown, there are two possible explanations. First, injection of fewer cells may result in less overall scattering within brain capillaries, thus only partially reducing blood flow, whereas higher cell doses may result in multiple MSCs entrapped within individual vessels, completely blocking blood flow and causing local ischemia. Alternatively, each single MSC may completely occlude an individual capillary, but, because of compensatory mechanisms, neighboring vessels provide sufficient coverage until a critical number of capillaries are blocked. Our observations about the consistent distribution of ischemic lacunar strokes within the white matter may be because of the specificity of cerebral circulation, with well-anastomosed and densely networked cortical circulation, and, as a consequence, a high resistance to local ischemia. In contrast, anatomically terminal deep brain arteries, in particular those supplying the white matter, are known for their less dense capillary network and are highly sensitive to local disruptions of the CBF.¹⁹ Intriguingly, the decrease in the dose of injected MSCs and the consequent prevention of stroke complications did not reduce CBF, as measured by laser Doppler flowmetry. This may be due to a relatively poor penetration of laser waves, which detect CBF primarily in the superficial, cortical brain regions, whereas strokes were observed in the deeper region of the corpus callosum. In fact, strokes were observed mostly within and in the vicinity of the corpus callosum, which resembles leukoaraiosis topographically—chronic brain ischemic disease, which is manifested predominantly as white matter malacia.²⁰

Even the most unfavorable experimental conditions in our studies did not result in a 100% stroke rate. This may be attributed to the variability of the vasculature among rats. Although stroke lesions after the injection of MSCs are likely because of cellular obstruction of capillaries, the reason for the occurrence of strokes after a rapid injection of PBS is less clear. The sudden mixing of a large volume of PBS and blood may cause microthrombosis, affecting the anatomically terminal arteries. However, these observations may not be directly translatable to the human setting, as there are differences in platelet function, with a stronger inhibition of aggregation in humans than in rats.²¹ Endothelial injury and the forced opening of the blood–brain barrier may be another reason for stroke occurrence, as stroke lesions may arise as a result of an acute elevation of the carotid pressure.²²

Although the slowest velocity we applied was 0.2 mL/minute, it might be that a further decrease of injection velocity incrementally over an extended period of time may allow safe delivery of a higher number of MSCs −2 × 10⁶ cells, or perhaps even more.

The application of laser Doppler flowmetry to monitor the CBF during intracarotid cell transplantation has been previously reported.¹⁰ It has been shown that the drop in CBF occurs during cell injection via a catheter, but does not occur when using a microneedle injection. In that study, 1 mL of cell suspension was injected via a catheter at a rate of 2 mL/minute and compared against the injection of only a 5 μL volume using a microneedle technique. Together with our observations, it is likely that the observed lacunar strokes are a result of the injected volume rather than the presence of cells. That study also reported that after cell infusion via a catheter, the CBF did not return to pretransplantation levels.¹⁰ In contrast, we consistently observed recovery of the CBF to pretransplantation levels within approximately 5 minutes. We also observed that fast infusion results in frequent stroke lesions despite a minor and brief reduction in CBF, whereas a profound reduction in the CBF after the injection of 1 × 10⁶ MSCs did not induce stroke. Thus, in our opinion, laser Doppler is not a reliable method for predicting stroke associated with intracarotid cell delivery.

In another study, injection of 2 × 10⁶ rat MSCs at a rate of 0.25 mL/minute²³ did not result in stroke lesions. The authors did not measure CBF and the cell size, and thus, the results are difficult to compare, although the injection rate was similar to that deemed safe in our studies. Injection of 2 × 10⁶ rat MSCs did not cause stroke, whereas we observed lesions with the same amount of hMSCs. This could be attributable to the smaller size of rat MSCs compared with hMSCs, similar to the measurements reported here showing that mGRPs are smaller than hGRPs. Other published reports on intraarterial cell delivery also did not include data on CBF or cell size, but an injection of up to 1.0 mL of cell suspension within 5 minutes also did not result in stroke complications.^{8, 24, 25, 26}

Conclusion

Intracarotid cell injections should be performed slowly, at a rate of 0.2 mL/minute in rats, to prevent the occurrence of stroke as a complication of the procedure. The intracarotid delivery of small cells, such as GRPs, with a size comparable to rat leukocytes, is safe, even if injecting relatively large cell doses (8 × 10⁶ cells). Injection of large cells, such as hMSCs, should be performed with caution, as the injection induces a sudden and profound decrease in CBF, and at doses of 2 × 10⁶ cells, introduces the risk of stroke. Laser Doppler measurements are not directly predictive of stroke complications after intracarotid cell delivery.

It should be realized, however, that the results of this study apply to preclinical rodent experiments, which may not be directly relevant to large animals or humans because of differences in vessel size, coagulation, and platelet function. Therefore, with the current experimental considerations as a guideline, independent safety studies should be performed in large animals as recommended by STEPS II and STAIR guidelines.

The authors declare no conflict of interest.