Tracking of stem cells in vivo for cardiovascular applications

Among the noninvasive imaging modalities, US continues to be the most commonly utilized for cardiac structure and function evaluations, in part due to its non-invasiveness, low cost, and portability. Moreover, the real-time interactivity of US and lack of ionizing radiation support the expansion of this imaging modality for stem cell delivery and tracking. However, until recently, ultrasound has been scarcely utilized in stem cell tracking because of the difficulty of attaching a long-lived ultrasound-visible label.

Tracking of stem cells via ultrasound can be done in combination with microbubble contrast agents, acoustically active liposomes, or perfluorocarbon nanoparticles. As previously stated, receptor labeling of stem cells is problematic because stem cells often lose markers as they differentiate as well as the inability to target exogenously delivered stem cells from native cells. One approach to overcome this problem has been spearheaded by Leong-Poi and co-workers, who have genetically modified endothelial progenitor cells (EPCs) to express the mouse H-2Kk protein [100]. Matrigel plugs containing H-2Kk-expressing EPCs were implanted subcutaneously in rats and subsequent ultrasound imaging demonstrated in vivo targeting of lipid microbubbles with anti-H-2Kk antibodies to the matrigel plugs whereas non-targeted microbubbles were not visible ultrasonically [100]. Beyond the disadvantage of genetic modification of stem cells, this promising approach also will suffer from problems of delivery of the microbubble to stem cells that are far from the vascular lumen. Because of this, most stem cell labeling using ultrasound has been targeted at adherence of stem cells to the vascular lumen or atherosclerotic plaque.

In a recent in vitro study, bifunctional, fluorescent and echogenic immunoliposomes (BF-ELIP) conjugated to anti-ICAM-1 antibodies for targeting to the atheroma were attached to bone marrow-derived CD34+ cells via anti-CD34 antibodies [101]. These BF-ELIP- labeled CD34+ bone marrow-derived stem cells were then incubated with freshly harvested porcine aortic tissue; ultrasound was then employed to enhance adhesion of the labeled cells to endothelium and enhance cell migration through the vessel wall rather than imaging [101].

In a similar vein, Toma et al. coated MSCs with cationic, gas-filled lipid microbubbles (mb-MSC), and used acoustic radiation generated by intravascular ultrasound at 1.7 MHz to encourage mb-MSCs delivered intra-aortically to adhere to the balloon-injured aorta in rabbits [102]. At 24 hours post-delivery, engrafted mb-MSCs remained localized to the luminal surface of the artery, but showed little migration beyond their original location on the luminal surface [102]. Recent refinements in this technique [103] are aimed at enhanced targeting of stem cells to specific sites of endothelial injury and additional mechanisms to encourage engraftment beyond the vessel itself.

Another approach is to use gold nanoparticles as acoustic reflectors. Nam et al. immobilized gold-labeled MSCs in a fibrin matrix and were able to perform longitudinal tracking [104]. Recently, internalization of polymer microbubbles by bone marrow-derived mesenchymal stem cells has been performed followed by in vitro ultrasound imaging [105]. The microbubbles, once internalized, remain acoustically active, and emit harmonics that are not exhibited by non-labeled stem cells. Unlike direct labeling with gold, if the cell died, one would anticipate that the microbubble shell integrity to degrade at which point detection of microbubble by ultrasound would cease as well.

Despite these advances, ultrasound tracking of stem cells is not widely employed as several significant hurdles must be overcome including: 1) the poor spatial resolution inherent with ultrasound imaging: 2) the intrinsic echogenicity of contrast agents preventing accurate cell quantification (agents cast acoustic shadows beneath the first unit of contrast encountered); 3) contrast agents can be diluted with each subsequent cell division; 4) agent stability is poor and higher stability may lead to uptake by phagocytic cells once cells die; 5) complicated acoustics result from uptake of the contrast agent into the cellular space; and 6) a limited field of view with the transthoracic, 2-dimensional-based technique thereby restricting access to many cardiac structures [82, 106]. In 2012 using lessons learned from high intensity focused ultrasound and enhance gene therapy transfection with ultrasound, Ziadloo et al. used pulsed focused ultrasound (pFUS) in a non-destructive manner to enhance bone marrow stromal stem cell homing and retention after intravenous administration to tissue treated with pFUS [107]. Thus, ultrasound may be used in a variety of ways to enhance cellular therapeutic engraftment and to assess cardiac function after delivery even if not directly used to track cells.

Cardiovascular magnetic resonance (CMR) typically interrogates the distribution of water within a subject. The high spatial resolution of CMR, as well as its capabilities for generating images with three-dimensional (3D) anatomical detail and lack of ionizing radiation makes CMR attractive for clinical application.

CMR-based cellular tracking has been performed using paramagnetic and superparamagnetic contrast agents as well as non-proton-based contrast techniques. Detection of labeled cells is related to magnetic field strength, labeling efficiency, cell numbers, relaxivity and spatial resolution [83]. Paramagnetic agents, such as gadolinium chelates and dysprosium, act locally on nearby protons to cause them to relax faster thereby shortening T1 relaxation. Gadolinium-based contrast agents, which are approved for intravascular administration, are used extensively in combination with T1-weighted imaging to increase signal in the vessels for magnetic resonance angiography, dynamic perfusion assessment in the heart, and viability assessment in the heart in delayed contrast-enhanced imaging. However, for direct stem cell labeling, paramagnetic agents have poor sensitivity due to the decreased effect on extracellular water once the contrast agent is intracytoplasmic [108].

Iron oxide contrast agents are commonly categorized based on particle size as: 1) superparamagnetic iron oxide (SPIO), typically between 30–150 nm in diameter; 2) ultrasmall superparamagnetic iron oxide (USPIO), typically <30 nm in diameter; and 3) micron sized particles of iron oxide (MPIOs) typically >1000 nm in diameter. Superparamagnetic contrast agents create substantial disturbances in the local magnetic field, which leads to a rapid dephasing of protons. Gradient echo techniques that cannot compensate for these dephasing artifacts will show hypointensities in the vicinity of iron oxide particles irrespective of whether the nanoparticles are internalized into the cell or not [83]. Thus, the sensitivity for tracking cells labeled with iron oxide nanoparticles is much higher than paramagnetic agents. To maximize the sensitivity to the magnetic susceptibility effects of USPIO/SPIO/MPIOs, T2*-weighted sequences are typically used. However, image interpretation may become difficult due to other endogenous sources of magnetic susceptibility, including hemorrhage and tissue interfaces, which also can create hypointensities on T2*-weighted images. Nonetheless, SPIOs have been used in a number of non-cardiac clinical trials for cell tracking outside the United States [109‐114].

Many preclinical studies have now been performed using iron oxide nanoparticles to help assess the optimal timing, stem cell type, dose, etc. In 2003, two studies heralded the use of iron oxide-labeled stem cells for cardiovascular cell delivery and tracking in large animals with CMR [115, 116]. The first study by Kraitchman et al.[115] used a technique called “magnetofection” to label stem cells with ferumoxides, a clinically approved SPIO for liver imaging. In this study, the uptake of the ferumoxides in non-phagocytic, bone marrow-derived MSCs was enhanced by the addition of a transfection agent, poly-L-lysine. The concept of combining SPIOs with transfection agents to enable rapid cell labeling without species or cell specificity developed by Frank and Bulte [117]. In this first study, CMR tracking of the SPIO-labeled MSCs after transmyocardial injection was performed in a swine, reperfused myocardial infarction model (Figure 3). In this study, ~30% of the injections of SPIO-labeled stem cells injected under X-ray fluoroscopic guidance were not visible under CMR indicating that perceived successful injections did not occur. Extrapolating these results to clinical trials with non-labeled cells, one could anticipate that there may be a variable response in clinical trials due to intermittent failure to deliver the stem cells. Thus, without cellular labeling, it is impossible to assess whether the therapeutic failed because of failure of the stem cells to be delivered, failure of the therapeutic to be retained in the heart or failure of the therapy itself due to a poor choice of injection dose, cell type, or timing of injection.

Hill et al. quickly followed with a similar study in swine using MPIO-labeled MSCs which were tracked up to three weeks after injection [116]. Building on this study, Dick et al. targeted MPIO-labeled MSCs to the infarct borders using a specialized active MR injection catheter and MR fluoroscopic imaging [118]. The advantage of the latter technique is that the success of the cell delivery could be immediately determined. Furthermore, using viability assessment with late gadolinium enhanced CMR [119], the cellular therapeutic could be specifically targeted to the peri-infarcted and infarcted tissue. In particular, Bulte and Kraitchman showed migration of stem cells in a reperfused dog infarction model over 8 weeks in the peri-infarction region [84]. SPIO-labeled MSCs were consistently removed when injected into normal myocardium whereas persistence of SPIO-labeled MSCs was noted in infarcted myocardium (Figure 4) [120].

Electromechanical mapping is another technique to target iron-labeled cellular therapy to viable myocardium and was first shown by Garot and colleagues for stem cell delivery in swine with infarcted myocardium [121]. In this study, injections in both normal and infarcted myocardium were all confirmed on CMR [121]. These results are in contrast to Kustermann and co-workers, who had difficulty differentiating the hypointensities created by USPIO-labeled cardiac progenitor cells from hypointensities due to the cryoablation or territory served by permanent coronary artery ligation to create damaged myocardium in mice. Presumably some of these issues were a result of the infarction model, small animal size, and small iron oxide particle size.

A variety of other investigators have studied iron oxide-labeled stem and progenitor cells with CMR [122‐127]. In a novel approach, Weber et al.[122] used magnetic beads designed for cell sorting to select endothelial progenitor cells, i.e., CD34 magnetic beads. However, the long-term retention of these magnetic beads on cells in vivo has never been fully investigated. In another murine study, Tallheden et al. injected SPIO-labeled embryonic stem cells under direct visualization in the anterior left ventricle and demonstrated hypointensities consistent with the labeled cells [123]. Subsequently, two additional groups demonstrated long-term tracking of SPIO-labeled stem cells in infarcted hearts [124, 125]. Stuckey et al. were able to track SPIO-labeled bone marrow-derived stromal cells up to 16 weeks post-infarction with CMR [124]. Ebert et al. not only tracked SPIO-labeled cardiac-differentiated mouse embryonic stem cells up to 28 days post-administration in a reperfused myocardial infarction, but also showed using CMR that declines in cardiac function were moderated by reductions in adverse remodeling [125]. Using MPIO-labeled amniotic fluid stem cells, Delo et al. were also able to demonstrate persistence of these cells up to 4 weeks after administration using CMR that was confirmed histologically [126]. Similarly, Chapon et al.[127] were able to demonstrate using CMR the persistence of hypointensities from rat bone marrow-derived stem cells labeled with a tat-peptide-USPIO nanoparticle [128] injected intramyocardially in both infarcted and sham operated rats at 6 weeks post-administration. A fluorescent label, FITC, in the tat-UPIO nanoparticle was used for histological validation. Despite using a 9.4 T CMR system, tracking of the hypointensities was more problematic, which the authors attributed to the smaller iron oxide nanoparticle used in this study.

Tracking of bone marrow-derived MSCs has been shown in a swine, reperfused infarction model by Hare and co-workers [129, 130]. In the first study [129], the appearance of increased subendocardial myocardium in the infarct zone in close proximity to stem cell injections was shown on multi-detector CT, and the presence of contracting myocytes was confirmed using tagged CMR [131, 132]. In a subsequent study [130], first pass contrast-enhanced CMR showed increased perfusion in the treated animals relative to controls prior to any functional benefit suggesting that MSCs may assist with angiogenesis and reduce apoptosis [130].

While these studies confirmed the presence of hypointensities on CMR corresponded to the exogenously labeled cells using histopathology, a number of studies have raised the concern that CMR hypointensities may reflect loss of the iron oxide nanoparticle from dead cells that either remains in the interstitial space or iron oxide nanoparticles that are taken up by phagocytic cells. In particular, Terrovitis et al. performed SPIO labeling of human and rat cardiac-derived stem cells (CDCs) that expressed beta-galactosidase and injected them intramyocardially in normal rats [133]. Persistence of hypointensities occurred for three weeks by CMR yet no CDCs were detected histologically. As expected, the area of the hypointensities decreased to a greater extent in the animals receiving xenogenic CDCs, presumably due to immunorejection, relative to the syngeneic CDCs whereas the hypointense area remained fairly constant from day 2 to day 21. Interestingly, the MR images in this study demonstrated a greater reduction in the hypointense area in the syngeneic animal compared to the xenogenic animal.

Schwaiger and colleagues performed a similar study in immunocompromised rats that received intramyocardial injections of iron oxide-labeled human endothelial progenitor cells (hEPCs) that were transfected with a PET reporter gene, the sodium iodide symporter (NIS) [134]. Both PET and CMR demonstrated the presence of the hEPCs at 24 hours post-injection. Hypointensities on CMR were still present at 72 hours and one week post-injection. On the other hand, PET imaging failed to detect any viable hEPCs at 1 week. Yet, immunohistochemistry at one week was unable to detect any of the exogenously hEPCs whereas staining for iron with Prussian Blue demonstrated co-localization of the iron with CD68+ cells, i.e., macrophages. While the authors note that PET imaging cannot detect less than 10,000 cells, the postmortem histology would suggest two things: 1. hEPCs did not survive even in an immune-compromised animal and 2. iron released from dying hEPCs is retained in macrophages that are still present at 1 week post-injection. In a related study, Li et al. performed serial imaging of SPIO-labeled human embryonic stem cells (hESCs) expressing a bioluminescence reporter gene that were injected into the hind limb of immunocompromised mice [135]. Undifferentiated ESCs are known for rapid proliferation and teratoma formation in vivo. As expected in this study, the development of teratoma resulted in an increased bioluminescence signal yet the CMR signal remained relatively constant. These reporter gene/CMR studies highlight two potential problems with direct labeling schemes: 1. that cell viability/proliferative capacity cannot be assessed and 2. that the label itself may become detached from the cell of interest. As such, direct labeling methods with SPIOs are best suited for targeting delivery and determining initial success rather than tracking cell viability. For long-term tracking, one must be cognizant that hypointensities may overestimate viable cells if free SPIOs or phagocytic cells, which scavenge the free SPIO, persist in the tissue. This issue is a problem in most cellular direct labeling methods.

Another issue with iron oxide nanoparticles is that the hypointensities created by the labeled cells compromise visualization of underlying tissue anatomy. Furthermore, other substances, such as air-tissue interfaces and metallic objects, e.g., stents, will create susceptibility artifacts that may mimic the hypointensities seen on T2*-weighted imaging of iron oxide-labeled cells. In an effort to overcome some of these issues, several techniques were developed based on a gradient dephasing technique by Seppenwoolde and co-workers [136] to passively track gadolinium-based markers, which cause local field inhomogeneities, in interventional devices. Using a twist on Seppenwoolde’s white marker technique and static dephasing theory of superparamagnetic particles explained by Bowen et al.[137], Mani et al. developed a slice-selective gradient echo technique for positive marker tracking of stem cells called GRASP [138‐140]. However, these dephasing techniques are very sensitive to imaging parameters, such as slice thickness and echo time. Alternative positive contrast SPIO imaging techniques have been developed that employ spectrally selective excitation/suppression [141‐143]. The spectrally-selective radiofrequency excitation and refocusing of off-resonant water frequencies proposed by Cunningham and colleagues [141] has been used in vivo to track embryonic stem cells implanted in the leg [135] and is best suited for spin echo techniques, which may limit applicability in the heart. An alternate off-resonant technique called inversion recovery with on-resonant water (IRON) saturates the water and fat peaks so that the off-resonant protons in close proximity to the SPIO-labeled stem cells are enhanced [142]. An added benefit to IRON imaging is that it can be used with either gradient echo or spin echo pulse sequences and can be applied across vendor platforms without any special pulse programming or post-processing of the images (Figure 5). Because these imaging techniques only preserve the susceptibility artifacts from the iron oxide-labeled cells, an additional anatomical reference image is needed. Alternatively Dahnke et al. have developed a post-processing method to determine the susceptibility gradient map from conventional gradient echo images to create positive contrast images of SPIO-labeled cell [144]. Thus, the standard anatomical image is obtained and no specialized pulse sequences are required. More recently, Garwood and coworkers have employed sweep imaging with Fourier transformation (SWIFT) for the detection of SPIO-labeled stem cells in the rat heart [145]. Using SWIFT, there is very little time between radiofrequency excitation and signal acquisition, which makes it well suited for imaging iron oxide labeling where the T2* relaxation time is very short. Interestingly, the magnitude image from SWIFT provides anatomical detail whereas the imaginary component provides a positive contrast image. However, the implementation of SWIFT is not available on commercial MR scanners nor is post-processing of the SWIFT images available currently. Another approach that has shown promise is to use ultrashort echo time (UTE) techniques available on clinical systems to acquire the T1 signature of iron oxide-labeled cells to create a positive contrast image [146, 147].

At one time, when clinically approved SPIOs, i.e., ferumoxides and ferucarbotran, were available, translation of SPIO-labeling clinical cardiovascular stem cell trials showed promise because of ease of labeling and low toxicity of the label if only to confirm stem cell delivery. However, these SPIOs have ceased to be manufactured for economic reasons [148]. Interest in USPIO labeling has been renewed with several groups demonstrating an off-label use of an FDA-approved injectable ferumoxytol for the treatment of iron deficiency anemia in adults with chronic kidney disease (Feraheme, AMAG Pharmaceuticals) to label a variety of cells [149, 150].

One potential solution to the problems associated with direct cell labeling with iron oxide nanoparticles would be to use a label that is rapidly removed if the cell dies, similar to microbubbles, which are under development for ultrasound cell labeling. In 2003, Ahrens et al. exogenously labeled dendritic cells with perfluorocarbons and performed in vivo tracking after direct tissue or intravenous injection in mice using Fluorine (19 F) MRI [151]. Because there is very little native fluorine in the body apart from the teeth, one can be exquisitely sensitive to fluorine labeled cells with the creation of hot-spot MR images similar to SPECT and PET imaging. In 2007, Partlow et al. labeled human umbilical cord progenitor cells with two different perfluorocarbon nanoparticles and demonstrated the ability to detect as few as 2000 labeled cells at 11.7 T in vivo and 1 million cells ex vivo at 1.5 T [152]. These cells were labeled with the perfluorocarbons by simple incubation without a transfection agent. Presumably, should the cells die, the perfluorocarbon would be excreted via the lungs rather than being retained in tissue. Subsequently, Barnett et al. labeled pancreatic islet cells with these same two perfluorocarbons and performed the first in vivo studies showing fluorine CMR in a clinical 3T scanner with retention of the labeled cells in the renal capsule of rabbits (Figure 6) [153]. While the scanner must have multinuclear capabilities and transmit/receive coils to detect the fluorine signal, this study demonstrates the potential for translation to clinical studies. Further aiding possible clinical translation is the availability of clinically approved perfluorocarbon agents that are used for echocardiography and as blood substitutes.

Another strategy that was originally designed to enhance survival of allogeneic cell therapies is microencapsulation. In the 1980s, alginate microencapsulation was developed by Lim and Sun [154] as a method to immunoisolate islet cells for transplantation in Type I diabetes mellitus. Typically alginate microencapsulation is performed in a multi-layer approach, e.g., alginate-poly-L-lysine-alginate (APA), to enhance the porosity of the capsule while retaining strength and biocompatibility. Thus, the APA microencapsulation technique enhances survival of allogeneic or xenogenic cells by restricting the passage of large molecules, e.g. immunoglobulins, while simultaneously allowing small molecules, e.g., oxygen, nutrients, cytokines, and waste products, to exchange across the membrane. Initially, Barnett and coworkers recognized that the poly-L-lysine moiety could be used for iron oxide labeling of the microcapsule and exploited this concept for tracking islet cell transplantation [155, 156]. More recently,this group has modified the formulation to incorporate perfluorocarbons for microcapsule tracking using fluorine MRI [157, 158]. One immediate advantage of this technique is that higher concentrations of the labeling agent can be added to the microcapsule without affecting cell viability, thus enhancing sensitivity. Using positive contrast techniques [142], this group has also demonstrated that one may be able to monitor the integrity of the iron oxide-labeled cell capsules [155]. Arifin et al. have shown that gadolinium chelates may also be incorporated into the alginate microcapsules when linked to a gold nanoparticle, which are visualized as hyperintense signals on T1-weighted MR images at 2-days after delivery [159]. However, similar to direct labeling techniques, this method of tracking cannot directly report the viability of the cell.

MRI reporter gene-based labeling entails the transfection of genetic material with plasmids or viral vectors to induce the cell to produce a specific receptor, protein, or enzyme that usually can be detected by the introduction of a reporter probe. One of the advantages of reporter gene imaging is that only live cells will produce the reporter gene product. Furthermore, if the reporter gene is not constitutively expressed but rather only by a specific promoter, there is the potential of imaging stem cell fate once differentiation down a specific lineage has occurred.

One of the earliest examples of MR reporter genes enabled imaging of gene expression. The approach by Louie et al. was to use a paramagnetic contrast agent whose access to water was blocked until cleaved by an enzyme [160]. Thus, cells expressing the enzyme would create an active form of the paramagnetic contrast agent and show increased intensity on T1-weighted images.

Another approach to MR reporter gene imaging techniques was to transduce cells to overexpress a native protein, ferritin, which is responsible for iron storage within the cell. Thus, cells overexpressing ferritin will accumulate more iron within the cell leading to a signal amplification that can be several folds greater than with direct receptor binding to the cell. Campan et al. recently demonstrated overexpression of the human ferritin heavy chain (hFTH) as a MRI reporter gene for in vivo tracking of swine cardiac progenitor cells in the infarcted rat heart [161]. Lentiviral-transduced cardiospheres (CSCs) overexpressing hFTH were injected intramyocardially at the perimeter of the infarct [161]. Iron accumulation in the rat hearts was followed up to 4 weeks using a multiecho, T2* gradient echo sequence on a 1.5 T clinical scanner [161]. While there is some concern that increased iron accumulation in the cell may be detrimental, CSC differentiation down multiple lineages still occurred [161], which would suggest that stem cell pluripotency was retained despite genetic manipulation and iron uptake. An MR reporter gene approach that does not rely on MR superparamagnetic or paramagnetic compounds has been explored in the brain. Gilad et al. used overexpression of lysine-rich residues as an endogenous contrast agent with increased amide residues, which could exchange protons with water residues [162]. Using chemical exchange saturation transfer (CEST) imaging [163], transfected cells could be distinguished from non-transfected cells in the brain [162]. Because paired images with and without radiofrequency irradiation are required for CEST, cardiac motion may be extremely challenging for implementation of this reporter gene approach.

One of the primary potential advantages of MR reporter gene imaging is that the reporter gene should be passed to the daughter cells and, thus, issues associated with label dilution during cell proliferation/division is markedly reduced. On the other hand, one must make sure that genetic expression as well as uptake of the reporter probe does not affect cell viability and function. In addition, there are concerns about the long-term expression of the foreign genetic material. Fortunately, silencing of the reporter gene often occurs over time, which may reduce fears of delivering a genetically altered cells to young cardiovascular disease patients.

CMR has several remaining hurdles to overcome as it relates to stem cell therapy for cardiac applications. For instance, quantification of labeled cell populations is challenging with the direct labeling schemes as there is dilution of intracellular markers with every cell division and there is potential for accumulation of iron particles within phagocytic cells that may lead to the production of false signals. Further, there is low sensitivity relative to radionuclide imaging, a lack of compatible devices for real-time delivery using CMR, MR contraindications in many cardiac patients due to metallic implants, the high expense of the CMR equipment, a high degree of acoustic noise, and relatively poor physiological monitoring for the acute cardiac patient. However, the lack of exposure to ionizing radiation to the patient, operator, and stem cells themselves in addition to the ability to determine myocardial viability, function, and perfusion in a regional manner have resulted in more frequent use of CMR to assess patients after stem cell delivery in cardiovascular clinical trials [5‐7, 9, 74, 75, 78, 79, 164, 165].

X-ray based imaging modalities include computed tomography (CT) imaging and X-ray fluoroscopy. The high spatial resolution and real-time interactivity are highly desirable attributes of these X-ray-based imaging modalities for clinical assessment and treatment of the heart. However, concerns about ionizing radiation dose and the limited ability to directly visualize soft tissue, such as the myocardium, are major drawbacks. Furthermore, most X-ray contrast agents are not amenable for direct cell labeling due to their high toxicity at relatively low doses.

Recently, Ricel et al. have develop a gold nanoparticles coated with poly-L-lysine for direct stem cell labeling [166]. To enhance biocompatibility, gold nanoparticles are typically coated with substances, such as polyethylene glycol (PEG), which in turn inhibits direct uptake of the nanoparticle by stem cells. Unlike MRI labeling compounds, gold nanoparticles have been shown to be actively expelled via exocytosis by cells [167]. Thus, loss of a gold label may occur even without cell division.

The development of MR-labeled alginate microcapsules has also been translated to X-ray-visible microcapsules. As with MRI, this allows the use of high concentrations of X-ray-visible contrast agents in the microcapsules without cellular toxicity for tracking using conventional X-ray fluoroscopic and computed tomography (CT) [156, 157, 168, 169]. Impregnation of APA microencapsulation with barium or bismuth sulfate allowed in vivo tracking of microcapsules in mice and rabbits using conventional X-ray fluoroscopy [156, 168]. These bariumsulfate-labeled microcapsules were also used to confirm delivery and retention of allogeneic MSCs in a rabbit model of peripheral arterial disease (Figure 7) [169‐171]. Single microcapsules could be seen in vitro using cone beam CT, although, in practice, several thousand microcapsules would be expected as a minimum to achieve a therapeutic effect. Another formulation using perfluoroctylbromide (PFOB) enables capsule tracking with X-ray, MRI, and ultrasound and may be useful for allowing delivery using X-ray fluoroscopic or ultrasound techniques while reducing radiation dose by using MR or ultrasound imaging for follow-up (Figure 8) [157].

One hurdle for application of microencapsulation techniques in the heart is the large size of the microcapsules, which is ~300-500 μm. Thus, this labeling method is not amenable to intravascular or transmyocardial delivery due to concern about vascular occlusion or induction of conduction abnormalities. As such, these techniques may be better suited for treatment outside the heart. Moreover, since the cells are trapped within the microcapsule, direct incorporation into the myocardium is unlikely. In 2011, Azene et al. demonstrated that an alternate delivery site for X-ray-visible microcapsules may be the pericardial space [173]. Using myocardial borders derived from a navigator- and cardiac-gated whole heart CMR at 1.5T fused with real-time X-ray fluoroscopy, barium sulfate impregnated microcapsules were delivered to the pericardial space in swine on a clinical angiographic system (Figure 9) [173]. These studies build upon X-ray fused with MRI (XFM) techniques [174, 175] developed in preclinical applications [176] for ultimate translation to pediatric and adult cardiac interventions [177, 178].

In the past 10 years, there have been a large number of preclinical studies and clinical trials that have used CMR to delivery, track, or determine the efficacy of stem cell therapy in the heart. While X-ray cell labeling techniques are not as mature at present for translation to clinical trials, the ability to fuse multiple image modalities, such as radionuclide imaging, CT, or MRI, with X-ray fluoroscopic imaging offers the ability to obtain the optimal interact interface for cell delivery with anatomical, functional, and metabolic information to better guide stem cell administration to the heart and peripheral vasculature.