Emerging roles for integrated imaging modalities in cardiovascular cell-based therapeutics: a clinical perspective

Abstract

Despite preclinical promise, the progress of cell-based therapy to clinical cardiovascular practice has been slowed by several challenges and uncertainties that have been highlighted by the conflicting results of human trials. Most telling has been the revelation that current strategies fall short of achieving sufficient retention and engraftment of cells to meet the ambitious objective of myocardial regeneration. This has sparked novel research into the refinement of cell biology and delivery to overcome these shortcomings. Within this context, molecular imaging has emerged as a valuable tool for providing noninvasive surveillance of cell fate in vivo. Direct and indirect labelling of cells can be coupled with clinically relevant imaging modalities, such as radionuclide single photon emission computed tomography and positron emission tomography, and magnetic resonance imaging, to assess their short- and long-term distributions, along with their viability, proliferation and functional interaction with the host myocardium. This review details the strengths and limitations of the different cell labelling and imaging techniques and their potential application to the clinical realm. We also consider the broader, multifaceted utility of imaging throughout the cell therapy process, providing a discussion of its considerable value during cell delivery and its importance during the evaluation of cardiac outcomes in clinical studies.

Introduction

Over the last decade, the use of progenitor cell therapies has emerged as an exciting option for the treatment of a range of cardiovascular diseases [1]. Preclinical small and large animal experiments have demonstrated that a diverse array of mature and immature cell types may confer benefit to myocardial function and perfusion after cardiac injury. This promise has translated into a steady stream of clinical research, beginning with early proof of principle studies [2, 3] and evolving to multicentre, placebo-controlled, randomised trials that are actively recruiting patients today. Several thousand individuals have now received unfractionated bone marrow cells (BMCs) [4–8], skeletal myoblasts (SkMs) [9], mesenchymal stromal/stem cells (MSCs) [10] or pro-angiogenic, endothelial progenitor cells (EPCs) [11] in studies investigating cellular treatment for acute and chronic myocardial infarction (MI), non-revascularisable ischaemic heart disease (IHD) and ischaemic and nonischaemic cardiomyopathy (NICM). Meta-analyses have concluded that cellular treatment for MI may lead to modest augmentation of left ventricular (LV) systolic function compared to controls (approximately 3% for absolute ejection fraction) [12, 13]. However, when analysed individually, trial results have been inconsistent [6, 14, 15]. Such discrepant observations are largely attributable to a lack of standardisation in the methodology employed by different investigative groups, relating to a variety of critical considerations, which can be divided into: (1) patient recruitment criteria, (2) cell product (type, preparation and dose), (3) timing and route of cell delivery, and (4) assessment of endpoints [16, 17].

As research efforts continue to address the biological limitations of different cell types, an important role has emerged for cardiac, cellular and molecular imaging to bring greater understanding and optimisation of the mechanics of cell delivery, the short- and long-term fate of implanted cells and their interactions with the host myocardium, as important interim outcomes after cell transfer. Here, we discuss the current and future utility of imaging as it relates to various key stages of the cell therapy process, focusing on image-based navigation of cell delivery to the myocardium and the noninvasive tracking of cell fate after administration, using clinically applicable strategies. Emphasis will be on the lessons learned from previous studies and where the focus should be placed in the future.

Imaging to assist cell delivery

A fundamental requirement of successful myocardial salvage is that sufficient viable cells reach their target sites soon after administration and are retained there to enable their long-term survival, engraftment, proliferation and function. Exogenous cells can be directed to the heart (1) systemically by peripheral venous injection, (2) regionally, by coronary arterial or venous infusion or (3) locally, by direct transepicardial, transendocardial or intrapericardial implantation. The choice of delivery method is likely influenced by the underlying disease process and cell type to be used, along with the expertise and resources of individual research groups. However, almost uniformly, studies have shown low rates of cell retention and engraftment across cell types, delivery methods and myocardial disease substrates, restricting the scope of benefit that can be achieved with current cell therapy strategies [18, 19].

By virtue of its practicality, safety and cost, fluoroscopically guided intracoronary infusion has been the commonest method used in clinical studies of cell therapy [6, 8, 14, 20], with the potential of cell distribution in the affected coronary artery territory. However, this approach may not be feasible in the setting of totally occluded coronary arteries and is hindered by direct washout of cells limiting first-pass retention and by the deleterious aggregation of adherent or large-sized cells (e.g. MSCs) within the coronary microvasculature [21–23]. Delivery by intramyocardial injection, either by open transepicardial or percutaneous, catheter-based transendocardial approach (Fig. 1a, b), targets specific areas of myocardium more directly. Although still hampered by significant early injectate loss [24], a direct injection approach appears to have advantages over systemic and intracoronary delivery with respect to cardiac cell retention [18], non-cardiac cell entrapment [22] and overall therapeutic effect [17, 25]. Cell retention and myocardial distribution are similar between the transendocardial and transepicardial strategies [26], although catheter-based injection is less invasive, increasing its broader clinical applicability and potentially making it amenable to repeated cell interventions [27]. It is in this context of transendocardial delivery, more so than for other delivery routes, that an important role has emerged for adjuvant imaging to guide the location of injection sites and determine the success of cell implantation in the myocardium [28].

Fig. 1.

Imaging navigation of transendocardial delivery. a Schematic displays an injection catheter passed across the aortic valve into the left ventricular cavity, with its needle tip extruded into the endomyocardium of the inferoposterior wall (inset). Purple areas of myocardium highlight two injection sites (image kindly provided by Biologics Delivery Systems Group, Cordis Corporation). b MyoStar™ injection catheter. The black arrow points to an inset of the catheter tip with extended 27 gauge needle and the red arrow to the adjustable injector thumb knob and Luer-lock fitting for connection to the injection syringe. c Conventional X-ray fluoroscopic guidance of catheter positioning near the left ventricular apex (adapted with permission from [98]). d, e Examples of electromechanical mapping images in a patient with anterior myocardial infarction, with linear local shortening (d) and unipolar voltage (e) maps shown in the anteroposterior projection. The red areas correspond to the infarct territory, with coupling of the reduced electrical signal and mechanical function over the anterior left ventricular wall. In this case, intramyocardial injections of cell therapy were administered in a peri-infarct distribution, as indicated by the brown circular markers (reproduced with permission from [33]). f–h MR guidance of catheter-based injection. The flexibility of the custom MR-compatible injection catheter (f) enables its manipulation to all regions of the endocardial surface. (g, h) Active catheters generate a high signal intensity for easy visualisation inside the ventricular cavity using real-time MR steady-state free precession imaging (reproduced with permission from [162])

Adjuvant imaging during catheter-based cell delivery

Most available injection catheter systems have been used with traditional biplanar X-ray fluoroscopy to visualise catheter manipulation and placement inside the LV cavity [24, 29–32] (Fig. 1c). This may be further assisted by pre-procedural imaging and planning with complementary modalities [e.g. echocardiography, single photon emission computed tomography (SPECT), positron emission tomography (PET) and magnetic resonance imaging (MRI)] to facilitate the selection of ischaemic or dysfunctional myocardial regions for cell delivery. Alternatively catheters have also been designed with sensor technology to allow their detection and navigation in real time with enhanced three-dimensional (3-D) precision. The most widely applied example of this is the MyoStar™ catheter (Fig. 1b) which is used in conjunction with the NOGA^® XP Cardiac Navigation System (Biologics Delivery Systems Group, Cordis Corporation, Diamond Bar, CA, USA) [33].

NOGA^® XP is a non-fluoroscopic, magnetic, electromechanical guidance technology that combines ultralow magnetic field sources (5 × 10⁻⁵ to 5 × 10⁻⁶ T) and location sensor-tipped catheter electrodes to accurately and reproducibly track a catheter’s trajectory inside the LV to within 1-mm distances. As endocardial sites are contacted and “sampled” by the mapping catheter, spatial, electrophysiological and mechanical data are acquired in real time to create 3-D, colour-coded reconstructions of the endoventricular surface. Electrical voltage amplitudes and mechanical contractility assessment (expressed as linear local shortening ratio) are used in combination to identify regional impairment of myocardial function, perfusion and/or viability. This enables the detection of non-viable scar and peri-infarct tissue in MI [34] (Fig. 1d, e), hibernating myocardium in chronic IHD and ischaemic cardiomyopathy [35, 36] and segmental fibrosis in NICM [37]. In turn, this information can be used to direct focused selection of target sites for cell delivery, with 3-D visualisation of injection density and distribution. Electromechanical navigation has thus been able to guide implantation of various cell types in large animal and clinical studies of chronic IHD [11, 38], acute MI [39] and most recently NICM [40]. It has also been used during follow-up to assess for improvement in regional electromechanical function after cell therapy, although this application has not been strictly validated [38, 41]. Over 50 clinical NOGA^® XP systems are currently in use [28]. Barriers to wider application include high cost and demand on operator expertise, training and accreditation [42]. Ongoing upgrades are designed to advance the technology by reducing mapping artefacts, shortening catheter response times, improving data accuracy and image quality and enabling stereotactic use [43, 44]. Non-contact electromechanical navigation is also under investigation for cell delivery, using the Endocardial Solutions (ESI)™ mapping system [45].

One shortcoming of catheter-based electromechanical mapping is its imperfect accuracy for sizing territories of ischaemia, infarction and fibrosis in the presence of severe LV dilatation [33]. Although more investigative, other catheter systems have undergone modifications to allow their coupling with real-time MR fluoroscopy (e.g. Stiletto™ [46] and MyoCath™ [47] catheters) and high-resolution 3-D echocardiography [48, 49]. MRI provides excellent 3-D anatomical and functional definition of the heart, along with high-resolution depiction and quantification of myocardial fibrosis and perfusion in both ischaemic and nonischaemic cardiac pathologies. The advent of ultrafast MRI technology and later generation, interactive scanners has opened up numerous possibilities for real-time MR fluoroscopy to be applied in interventional cardiovascular practice including the targeted delivery of endovascular and intramyocardial injectates (e.g. cells, genes, drugs) [50, 51]. Various scanning systems have been created for MRI-based interventions, including the hybrid XMR system which integrates real-time X-ray and non-ionising MR fluoroscopy for flexible, complementary imaging [52, 53]. Visualisation of catheters can be achieved by passive or active tracking [50]. Passive devices are detected on MR images as a signal void or susceptibility artefact, which can be enhanced by coating the catheter shaft or tip with paramagnetic markers (e.g. dysprosium oxide, gadolinium oxide, ferrite admixtures) and by administering blood pool contrast medium. An advantage of this type of tracking is that it can be performed with conventional MR scanners, without adjustments to hardware or software or the need for image post-processing. However, these catheters are prone to safety hazards during MR fluoroscopy due to ferromagnetic attractive forces or overheating during radiofrequency transmission [54]. By comparison, active tracking requires catheters to be equipped with receiving microcoils along their shaft and/or tip that interface with the MRI scanner to provide signals within the native anatomy. This usually necessitates specialised hardware and post-processing technology to superimpose the image of the coil on the road map image of the heart. Such catheters have been used to perform MRI-guided intramyocardial injections in large animal studies with good precision and safety [46, 55, 56] (Fig. 1f–h).

Despite the promise of initial preclinical validation, there has been limited ongoing research or manufacturing interest to develop US Food and Drug Administration (FDA)-approved, MRI-compatible catheters for transendocardial intervention. Concerns also surround the ability to monitor acute cardiac patients in a high magnetic field environment during invasive procedures, and additional expertise would be required for the interventional cardiologist to interpret the graphical interfaces during real-time MRI acquisition. In the short term, the role for cardiac MRI during cell delivery is therefore likely to be as an adjunct to other types of navigation imaging. Custom-made software may allow previously acquired MR images to be merged with real-time X-ray fluoroscopy [57] or electromechanical mapping [58]. This practice of bimodal imaging enhances 3-D anatomical and tissue characterisation so that injections can be avoided in areas of thinned or non-viable myocardium.

In summary, biplanar X-ray fluoroscopy and electromechanical mapping currently remain the most established tools for guiding transendocardial cell delivery and have been utilised in a variety of completed clinical studies (Table 1), as well as numerous ones that are ongoing (www.clinicaltrials.gov). Technical modifications and the integration of adjuvant imaging modalities are expected to further strengthen the role of real-time imaging in navigated cell delivery. It is also likely that imaging will facilitate the emergence of new approaches to cell delivery, such as percutaneous intrapericardial injection, which has recently been performed in pigs under guidance by X-ray fluoroscopy together with intravascular ultrasound [59] or MRI [60].

Table 1.

Clinical studies using image navigation for transendocardial cell delivery

Study authors (study name)	Catheter	Disease	Design	No. of patients	Cell type
(A) Electromechanical mapping
Krause et al. [39]	MyoStar™	Acute MI	Non-controlled	20	BM MNC
Gyöngyösi et al. [155] (MYSTARa)	MyoStar™	Acute and chronic MI	Randomised time comparison	30 early, 30 late	BM MNC
Briguori et al. [156]	MyoStar™	Chronic IHD	Non-controlled	10	BM MNC
Beeres et al. [157]	MyoStar™	Chronic IHD	Non-controlled	25	BM MNC
Fuchs et al. [158]	MyoStar™	Chronic IHD	Non-controlled	27	BMC
Losordo et al. [11]	MyoStar™	Chronic IHD	RDBPCT dose-escalation	18 Rx, 6 placebo	G-CSF + CD34+ cells
Tse et al. [159] (PROTECT-CAD)	MyoStar™	Chronic IHD	RBPCT dose-comparison	19 Rx, 9 placebo	BM MNC
van Ramshorst et al. [160]	MyoStar™	Chronic IHD	RDBPCT	25 Rx, 24 placebo	BM MNC
Perin et al. [38]	MyoStar™	Ischaemic CMP	NRCT	11 Rx, 9 control	BM MNC
Dib et al. [161] (CAuSMIC)	MyoStar™	Chronic MI/ischaemic CMP	RPCT dose-escalation	12 Rx, 11 placebo	SkM
(B) X-ray fluoroscopy
Ince et al. [29]	MyoCath™	Chronic MI/ischaemic CMP	Case-controlled	6 Rx, 6 control	SkM
Smits et al. [30]b	MyoStar™/MyoCath™	Chronic MI/ischaemic CMP	Non-controlled	15	SkM
de la Fuente et al. [31]	Helix™	Chronic MI/ischaemic CMP	Non-controlled	10	BM MNC
Duckers et al. [32] (SEISMIC)	MyoCath™	Ischaemic CMP	RCT	26 Rx, 14 control	SkM

BM bone marrow, BMC bone marrow cells, CMP cardiomyopathy, G-CSF granulocyte colony-stimulating factor, IHD ischaemic heart disease, MI myocardial infarct, MNC mononuclear cells, (N)RCT (non-)randomised controlled trial, R(DB)PCT randomised (double-blinded) placebo-controlled trial, Rx cell treatment, SkM skeletal myoblasts

MyoCath™ (Bioheart Inc., Sunrise, FL, USA) and Helix™ (Biocardia, Inc., South San Francisco, CA, USA)

^aMYSTAR study compared timing of delivery for combined intramyocardial and intracoronary routes

^bBoth types of catheters were used in this study, with their respective imaging modality for navigation

Imaging to assess cell fate

The ability to accurately trace cell fate is of utmost importance if key questions are to be addressed regarding optimal cell type and dose, efficiency of cell delivery, cell biodistribution and engraftment and the mechanisms by which cells may impart myocardial restoration (differentiation versus paracrine effects). Noninvasive tracking is therefore necessary to temporally monitor the presence of cells in intact subjects and to elucidate their kinetics and biological interactions with recipient myocardium over time.

In order to be visually distinguishable from endogenous tissue, exogenous cells must first be labelled (for direct or indirect monitoring) prior to their administration. Ideally, techniques used for cell labelling and imaging should be non-toxic to both the cells and the recipient organ(s) and possess a good balance of spatial resolution and cell detection sensitivity. Furthermore, they should have low enough radiation and contrast exposure to enable serial monitoring at multiple time points, high specificity so that signal can be interpreted as coming exclusively from viable, labelled cells and good quantitative accuracy to provide a reproducible measure of actual cell numbers.

Focusing on clinically translatable modalities, cell labelling is usually performed either (1) directly, which is best suited to the assessment of short-term retention and distribution of cells or (2) indirectly by reporter gene transfer, which may enable longer-term monitoring of cell engraftment, viability and interaction with the myocardium (Fig. 2).

Fig. 2.

Direct and indirect labelling for cell detection. Schematic depictions of the strategies used for a direct labelling and b indirect reporter gene labelling of cells prior to their administration, with examples of their imaging using MRI, SPECT and PET. c Delayed enhancement MRI showing infarcted myocardium (hyperintense) containing hypointense lesions from iron-labelled MSCs (arrows) which were injected 24 h earlier (adapted with permission from [76]). d Combined PET/CT image showing in vivo detection of ¹⁸F-fluorodeoxyglucose (FDG)-labelled progenitor cells within the inferolateral territory of the left ventricle after intracoronary delivery (borrowed with permission from [68]). e SPECT-based imaging of the biodistribution of ^99mTc-hexamethylpropylenamineoxine (HMPAO)-labelled BM progenitor cells 1 h after delivery to a patient with chronic ischaemic cardiomyopathy. Anterior view of chest and upper abdomen is shown with black indicating no uptake and blue-red-yellow-white showing the gradient of increasing signal (reproduced with permission from [66]). f Fusion image of MRI (grey scale) and ¹⁸F-FHBG PET (hot scale) showing accumulation of tracer in porcine myocardium at two sites where MSCs transfected with HSV-tk were injected 8 h earlier. g An example of hybrid ¹⁸F-FHBG PET/CT imaging from the same study confirmed the localisation of injected cells in the anterior left ventricular wall (both images are adapted with permission from [122]). CMV cytomegalovirus, HSV-tk herpes simplex virus thymidine kinase, SPIO superparamagnetic iron oxide, ¹⁸F-FHBG(P) 9-(4-¹⁸F-fluoro-3-[hydroxymethyl] butyl)guanine (phosphate)

Assessment of short-term cell fate

The most commonly used approach to label cells directly is by incubating them in vitro with magnetic contrast agents for MRI surveillance [61, 62] or radionuclide agents for detection with SPECT [63–66] or PET [19, 67, 68].

Magnetic resonance imaging

Cardiac MRI has several qualities that make it a good candidate for serial imaging of individual subjects, as required during longitudinal assessment of cell therapy efficacy [69]. In what relates to the monitoring of administered cells, direct labelling for MRI has been mostly achieved with magnetic nanoparticles (MNP), which are typically superparamagnetic but can also be synthesised to be ferromagnetic or paramagnetic. Superparamagnetic MNP induce strong magnetic field disturbances that reduce T2* relaxation time and create hypointense signal voids on T2- and T2*-weighted images. They are usually composed of an iron oxide core (magnetite and/or maghemite) measuring 3–5 mm in diameter, with a polymeric or polysaccharide coating (e.g. dextran), which maintains their solubility and reduces their agglomeration. This structure renders them biodegradable via iron metabolic pathways and largely biocompatible. With physical and chemical modifications, several generations of MNP have now been synthesised. Early generation MNP comprised superparamagnetic iron oxide particles (SPIO, 60–150 nm in diameter), such as the ferumoxide, Feridex (Advanced Magnetics, Cambridge, MA, USA) and micron-sized iron oxide particles (MPIO, 0.7–1.6 µm) [70]. SPIO contain relatively thin dextran coats and tend to form polycrystalline clusters, allowing their rapid clearance from the circulation by the reticuloendothelial system. Subsequently smaller MNP have been developed, including citrate-coated ultra small paramagnetic iron oxides (USPIO, 10–40 nm) and monocrystalline iron oxide nanoparticles (MION 10–30 nm). As a result of their extensive polymer coating, these nano-sized MNP remain monodisperse in solution, with longer circulatory half-lives. They also have higher relaxivities than first-generation MNP, allowing deeper penetration into tissue spaces and conferring higher sensitivity. Highly stable, cross-linked derivatives of MION, known as CLIO, have become particularly attractive for targeted molecular imaging [71] and can be conjugated to fluorochromes (e.g. rhodamine B isothiocyanate) for multimodal detection by MRI and optical imaging techniques [72].

Iron oxide-labelled cells have been successfully imaged in both small [73–75] and large animal cardiac models [61, 76, 77] using clinical MR scanners (1.5–3 T). In this context, transplanted cells (≥10⁵ cells) have been visualised in the initial hours to days after implantation, providing verification of the success of their delivery and their precise location in the myocardium (e.g. relative to tissue scar). Cells on biomaterial scaffolds, a subject of continued investigation to enhance engraftment in recipient myocardium, have also been imaged [78, 79]. Although SPIO-based labelling has regulatory approval for the imaging of liver tumours [80] and lymph node metastases [81], the cessation of ferumoxide production in the USA and Europe poses a setback to clinical translation. As such, MRI tracking of cell therapy has not yet been adopted in human studies of cardiovascular disease, although there are encouraging precedents for its use in patients with melanoma [82] and brain trauma [83].

Despite their practical simplicity, iron oxide nanoparticles have several shortcomings including their potential for cytotoxicity [84]. Although they do not appear to negatively effect cell viability or proliferation at the concentrations required for cardiac MRI [85, 86], impairment of other cellular properties has been documented, including reduced migration of EPCs [87] and diminished colony formation, migration and chondrogenic differentiation of MSCs [88, 89]. Like all direct labels, iron oxides also have limited utility for long-term tracking as they are not self-replicable and become diluted after cell fragmentation, fusion, division or migration [90]. Moreover, their physical detachment after cell death and subsequent ingestion by tissue-resident macrophages can result in persistent, nonspecific signal that may be misinterpreted as cell engraftment [75, 84, 91]. Quantification of cell number can also be hindered by the “blooming effect” of iron oxides and by signal void confounders such as myocardial haemorrhage or microvascular obstruction after MI [92]. Positive contrast techniques, specialised imaging sequences and post-processing methods may overcome some of these limitations [93, 94]. Alternative approaches are to use hyperintense signals on T1-weighted images to track cells labelled with paramagnetic contrast agents, such as manganese chloride compounds, traditional gadolinium chelates, or novel, high-sensitivity compounds which may have better safety profiles (e.g. gadolinium-containing carbon nanocapsules [95], gadofluorine M-Cy3 [96]). Cell tracking has also been performed using non-proton fluorine MRI (¹⁹F-MRI), which has a higher signal to noise ratio compared to standard proton-based MRI, allowing it to sensitively detect perfluorocarbon-labelled cells as “hot” spots [97]. Although experimental at present, many of these advances in MRI tracking may be translatable to clinical studies in the future.

Radionuclide imaging

Direct labelling for cell tracking by SPECT or PET can be performed by incubating cells with radioactive tracers, many of which are already in clinical use. Examples include ¹¹¹In-oxine and ^99mTc-hexamethylpropylenamineoxine (HMPAO) for SPECT and ¹⁸F-fluorodeoxyglucose (FDG) and ⁶⁴Cu-pyruvaldehyde-bis(N4-methylthiosemicarbazone) (PTSM) for PET. Cell detection sensitivity is higher with PET (femtomolar) and SPECT (nanomolar) compared to MRI (micromolar). This advantage is offset by their exposure to ionising radiation, as well as their lower spatial resolution (millimetres compared to micrometres for MRI), which limits the anatomical detail with which cells can be localised to the myocardium. However, this can be improved by performing hybrid imaging whereby radionuclide images are integrated with those from computed tomography (CT) or MRI [26, 98]. PET, in particular, allows accurate quantification of cell number by measuring retention as a percentage of total injected dose [99]. As opposed to PET, SPECT-based cell labelling is able to detect simultaneous signals of different energies (80–250 keV) by varying the detection windows, enabling its dual-purpose application for cell tracking concurrent with ²⁰¹Tl- or ^99mTc-based perfusion imaging.

The physical half-lives of radiotracers partly determine the length of follow-up possible. In the case of ¹⁸F-FDG (half-life of 109 min), monitoring is generally focused to the first few hours after cell delivery [67], although this may be extended with higher sensitivity scanners (e.g. LSO and GSO detectors). Surveillance for a few days is possible with PET-based tracking of ⁶⁴Cu- PTSM (half-life of 12 h) and SPECT imaging of ¹¹¹In compounds (half-life of 2.8 days). Long-term monitoring of radiolabelled cells is also prevented by the aforementioned, inherent limitations of direct labelling that result in signal dilution over time. Biological half-life, labelling efficiency and cytotoxicity are all important considerations when applying different radionuclide labels to specific cell types. For example, cell uptake of ¹¹¹In compounds is modest and their emission of high-energy electrons has resulted in cytotoxic effects, such as reduced viability and colony-forming ability of haematopoietic progenitor cells [100, 101] and compromised viability, proliferation, migratory capacity and metabolic integrity of MSCs [102–104]. Better cell tolerability has been shown with the PET tracer ⁶⁴Cu-PTSM [105].

Cell imaging with SPECT or PET has been implemented in a number of preclinical studies to address key questions concerning the efficiency of cell retention for different cell types and delivery techniques [18, 26, 68, 106, 107]. Early rodent studies shed light on the initial homing of cells to the infarcted myocardium after systemic delivery and the temporal profile of their distribution to the lungs followed by the liver, kidneys and spleen [100, 106]. In large animal studies of MI, radionuclide tracking has been used to demonstrate better myocardial retention of cells, with less pulmonary entrapment, after direct intramyocardial implantation compared to peripheral or coronary vascular infusion [18, 63]. Recently, serial SPECT/CT imaging was used to show that retention rates were similar between surgical transepicardial and percutaneous transendocardial injection of ¹¹¹In-tropolone-labelled EPCs in dogs [26]. Valuable lessons were also learned from a study in which ¹⁸F-FDG-labelled cells were imaged with dynamic PET/ CT to compare two different strategies of intracoronary infusion [68]. After cell delivery with repeated cycles of the stop-flow infusion technique (common in many clinical trials), myocardial ¹⁸F-FDG signal (as a surrogate for the presence of cells) was transiently higher during balloon occlusion, before falling sharply during balloon deflation. In contrast, first-pass clearance of cells was observed to be more gradual, with higher resultant myocardial retention, when cells were delivered as a single, high-concentration bolus.

Unlike MRI, radionuclide detection of directly labelled cells has also been applied in the realm of clinical cardiovascular studies (Table 2). Hofmann et al. described a small series of MI patients, in whom 3-D PET imaging was performed within 90 min after intracoronary or intravenous infusion of ¹⁸F-FDG-labelled cells (unselected BMCs or CD34-enriched cells) [19]. Cardiac cell signal was not observed after systemic administration of unselected cells, but was present at very low levels (1.3–2.6%) in the infarct and border regions after intracoronary delivery. Retention was augmented considerably for CD34⁺ cells which distributed selectively to the border zone (14–39%). Similar data have been reported by other radionuclide studies in patients with acute or chronic MI [64, 66, 67, 108]. Further insights were also obtained from a study in which SPECT was used to monitor the retention of ¹¹¹In-labelled pro-angiogenic cells after intracoronary injection in patients with different aged infarcts [109]. Cardiac signal activity was highly variable, averaging 6.9% (range 1–19%) 1 h after cell transfer, before declining to 2% after 3–4 days. Retention was highest in patients with recent MI (<14 days old), progressively diminishing in those with intermediate (up to 1 year old) or chronic (>1 year) infarcts.

Table 2.

Clinical studies of cell labelling and tracking

Study	Disease	No. of patients	Cell type/ delivery route	Cell label	Imaging modality	Time	Results
Hofmann et al. [19]	Acute MI	3 (IC BMC), 3 (IV BMC), 3 (IC CD34+)	Unf BMCs or CD34+ cells/IC or IV	18F-FDG	PET	50–75 min	Background signal only with IV. Augmented signal with CD34+ cells. Signal in border ± MI zone
Kang et al. [67]	Acute MI	17 (IC), 3 (IV)	PB MNCs/IC or IV	18F-FDG	PET/CT	2 h, 20 h (n=1)	No MI signal after IV. 1.5% retention in heart after IC
Silva et al. [64]	Acute MI	14 (IC), 10 (RICV), 6 (control)	BM MNCs/IC or RICV	99mTc-HMPAO	SPECT	4 h, 24 h	Retention higher with antegrade IC delivery route
Penicka et al. [65]	MI	5 (acute MI), 5 (chronic MI)	BM MNCs/IC	99mTc-HMPAO	SPECT	2 h, 20 h	Cardiac signal in 5/5 acute and 4/5 chronic patients at 2 h and in 3/5 acute and 0/5 chronic patients at 20 h
Schächinger et al. [109]	MI	8 (acute MI), 4 (inter MI), 5 (chronic MI)	PB MNCs/IC	111In-oxine	Whole-body scintigraphy	1 h, 24 h, 3–4 days	Signal highest after 1 h and in acute MI group
Goussetis et al. [66]	Chronic MI/ischaemic CMP	8	BM CD133+ or CD34+ cells/IC	99mTc-HMPAO	SPECT	1 h (n=8), 24 h (n=4)	Cardiac retention 9.2% at 1 h and 6.8% at 24 h. Extracardiac signal in spleen and liver
Dedobbeleer et al. [108]	Chronic MI	7	PB CD34+ cells/IC	18F-FDG	PET/CT	1 h	3.2% retention at MI borders. Higher signal in liver, spleen, BM

All of the above studies involved direct cell labelling with radionuclide labels. Time column indicates the time after cell delivery that imaging was performed

CT computed tomography, IC intracoronary, inter intermediate aged, IV intravenous, PB peripheral blood, PET positron emission tomography, RICV retrograde intracoronary vein, SPECT single photon emission computed tomography, Unf unfractionated, ¹⁸F-FDG ¹⁸F-fluorodeoxyglucose, ^99mTc-HMPAO ^99mTc-hexamethylpropylenamineoxine. Other abbreviations are defined in Table 1

By consistently revealing the modest extent to which cells are retained in the heart, as well as their distribution to other organs, the above-mentioned tracking studies have stimulated vigorous research to optimise cell delivery and biology in order to improve cell engraftment and treatment outcomes [99, 110]. Their findings also provide useful context to help contemplate the inconsistencies between clinical trial results, the merits of systemic cell delivery [10] and the relative value of using unfractionated BMCs [6, 14], compared to enriched cell populations [11]. These studies also highlight the individual variability of cell therapy responses and the need for future treatment regimes to be carefully tailored to each patient.

Assessment of long-term cell fate

After successful delivery and early retention of cells in the heart, it is desirable to monitor their long-term survival and functionality, which may be achieved by indirect cell labelling using reporter gene transfer. In this imaging strategy, cells are engineered to produce a non-native or overexpressed enzyme, receptor or protein and when this protein interacts with an exogenously given substrate it results in signal that can be used to distinguish implanted cells from endogenous cells with high specificity. Unlike direct labelling techniques, the signal from reporter gene labels is only produced by viable cells with intact metabolic and synthetic function. Moreover, when the transgene has been permanently incorporated into the genomic DNA (i.e. stable transfection, as with retroviral or lentiviral vectors), it is passed on to daughter cell progeny during cell replication, avoiding signal dilution due to cell division. Reporter gene labelling is perhaps most attractive for highly proliferative cells (e.g. embryonic stem cells [111] and induced pluripotent stem cells [112]) where relatively few cells require transfection before clonal expansion.

Thus far, reporter gene strategies to track stem cell survival have been mainly based on optical bioluminescence imaging (BLI) using luciferase reporter genes (e.g. firefly luciferase, Renilla luciferase) in small animal studies [84, 111, 113, 114]. Expression of these enzymes allows the cells to oxidise specific probes (e.g. d-luciferin, coelenterazine), resulting in their emission of light photons, which can be detected by ultrasensitive charge-coupled device (CCD) cameras. Despite its many advantages for cell tracking in rodents (low cost, non-toxicity, high sensitivity), BLI suffers from low spatial resolution and a lack of tissue depth penetration, precluding its use in human subjects. However, reporter genes can also be adapted to clinical imaging modalities, most notably PET [115], but also SPECT [116] and MRI [117, 118].

Previous publications have described in detail the different types of reporter genes available for radionuclide detection [119–121]. In brief, these are based on either (1) intracellular enzymes which can trap the probe (e.g. herpes simplex virus thymidine kinase, HSV-tk) [111, 122], (2) cell membrane transporter proteins which enable probe uptake (e.g. sodium-iodide symporter, NIS) [116, 123] or (3) cell membrane receptors which bind directly to the probe (e.g. mutant dopamine D₂ receptors) [124]. Compared to SPECT, PET has significant flexibility for the production of specific probes to detect different processes (reporter genes in this case) in the living subject, allowing researchers to first identify the molecule to be imaged and then design a specific probe that will target that molecule. On the other side, SPECT imaging provides the possibility to investigate more than one signal simultaneously.

In vivo applications have also been described using reporter genes specially designed for targeted molecular MRI [117, 118]. These are based on the production of various proteins, such as enzymes that block water (proton) exchange, surface receptors that bind MR contrast agents or intracellular metalloproteins involved with iron metabolism (ferritin, transferrin receptor, tyrosinase) [117, 118, 125]. The ferritin gene is especially attractive as it greatly facilitates intracellular sequestration of iron (Fe³⁺) resulting in T2/T2* shortening without the need for concomitant exogenous contrast agents. However, despite initial enthusiasm, MR imaging of reporter gene activity is hindered by higher background signal and lower sensitivity compared to radionuclide modalities and thus is not widely available or feasible for clinical translation in the foreseeable future.

So far, reporter gene monitoring has been performed predominantly in rodent models of cardiovascular disease [112, 115] and experience in large animal studies has been relatively limited [122, 126, 127]. In a porcine model of MI, in vivo and ex vivo techniques were used to track MSCs that had been labelled with triple fusion reporter genes [Renilla luciferase, red fluorescence protein (RFP), herpes simplex truncated thymidine kinase] and delivered by NOGA-guided transendocardial injection [122]. Engraftment of cells was detected by ¹⁸F-FHBG PET and their localisation was verified by fusing these images with CT or MRI. Hybrid images revealed diffuse myocardial distribution of cells in the early stages after focal injection, with diminution of cardiac signal and uptake in other tissues after 7 days. This elegantly demonstrated that viable cells continue to be lost from the myocardium after their initial retention, probably through both lymphatic and circulatory migration to other organs. In another recent porcine study, PET/CT imaging demonstrated the value of using bioscaf-folds to enhance retention of human MSCs that had been adenovirally transfected with HSV1-sr39tk [127]. At this stage, there is only one non-cardiac clinical report for reporter gene imaging of cells, in which PET was used to monitor HSV1-tk transfected cytolytic T cells in a patient with glioma [128].

Although reporter gene tracking offers a promising strategy for the long-term assessment of cellular engraftment, uncertainties remain relating to the safety of different viral vectors, the deleterious effects of genetic modification on cell integrity and function, the possibility of non-integrative (episomal) gene transfer resulting in unreliable or temporary signal and the diminution of transgene expression over time.

Multimodal cell tracking

Selection of the most appropriate labelling-imaging combination requires careful consideration of the respective advantages and disadvantages of each strategy (Table 3) and its ability to address the specific question in mind. It is also evident that in order to achieve comprehensive assessment of both short- and long-term cell fate, a multimodal or hybrid approach to cell tracking may be helpful, by integrating different techniques with complementary strengths [129, 130]. In terms of imaging modality, the higher spatial resolution of MRI is well suited to localising cell distribution in the myocardium, whereas the superior sensitivity and quantitative capabilities of radionuclide techniques (PET or SPECT) are more useful for determining cell number. As discussed before, reporter gene tracking is currently most advanced with PET and SPECT, thereby providing the best option for the long-term study of cell biology and fate. Fusion of PET or SPECT images with higher resolution spatial techniques (e.g. CT, MRI) has enabled radionuclide cell signals to be interpreted more precisely in their anatomical context [122]. Although this can be achieved by offline coregistration using fiducial markers in the field of view and specialised shape/pattern recognition software [131, 132], coregistration may be affected by differences in slice thickness between the modalities (e.g. PET, CT) and movement of the patient (or organ), while signal resolution may also be dependent on the size of markers. These issues are more avoidable with clinically available, integrated, hybrid systems. While current radionuclide/CT scanners obtain their respective images sequentially, MRI-compatible PET scanners have now been designed to allow data from both modalities to be obtained simultaneously without significant cross-interference of image quality [133]. Recently, Siemens Medical Solutions announced the development of a whole-body MR/PET system (Biograph™ mMR) for human patients.

Table 3.

Comparison of imaging modalities for cell tracking

Monitoring strategy	Spatial resolution	Cell detection sensitivity
Direct labelling
SPECT	++/+++	+++
PET	++/+++	+++
MRI	++++	++
Indirect labelling (reporter genes)
SPECT	++/+++	+++
PET	++/+++	+++
MRI	++++	Unknown

Comparison of the spatial resolution and cell detection sensitivity of the three main imaging modalities that are translatable to cell tracking in clinical practice. Scale is semi-quantitative: + (least) to ++++ (most)

MRI magnetic resonance imaging, PET positron emission tomography, SPECT single photon emission computed tomography

Future strategies for cell labelling and imaging

In addition to MRI and radionuclide techniques, echocardiography may also be an option for cellular imaging in the future, probably through contrast enhancement by microbubble targeting. An example of this principle was provided in a recent rodent experiment, in which EPCs were transfected to express a cell surface marker protein (H-2Kk) that could be detected on ultrasonography by using microbubbles coated with a specific monoclonal antibody [134]. Another interesting possibility that has been evaluated to track transplanted pancreatic islet cells in vivo, but may have a broader application, is cell microencapsulation using semi-permeable capsules (e.g. alginate) that are embedded with contrast agents [e.g. bismuth sulphate, perfluorocarbon (PFCs)-hydrocarbons] [135, 136]. These contrasts may serve a dual function in that they seem to improve cell viability and function, while also rendering cells visible to various imaging modalities, including X-ray fluoroscopy, CT, ultrasound and ¹⁹F-MRI. In the future, it is likely that more novel imaging strategies will be added to the armamentarium for the monitoring of cell therapies.

Integrated imaging of cell therapy in clinical studies

As depicted in Fig. 3, there is extensive scope for imaging to facilitate virtually every stage of the cell therapy process in patients with cardiovascular disease, from patient selection through to cell delivery, assessment of cell fate and follow-up of therapeutic outcome. One of the fundamental roles for imaging in clinical studies is to help characterise which patient and disease cohorts are most likely to respond to specific treatment strategies [137]. This requires rigorous baseline and follow-up evaluation of cardiac parameters that are relevant to disease context and are appropriate predictors of clinical outcome. Thus far, the benefit from autologous BM cell therapy after MI appears to be largely restricted to patients with the most severely impaired systolic function, as measured by LV ejection fraction [12]. It is therefore highly desirable for investigators to have access to objective, operator-independent modalities that can provide accurate and reproducible quantification of global ejection fraction, especially in the presence of LV dilatation and dysfunction [138]. To this end, a large number of clinical trials have favoured the use of MRI [70], as well as radionuclide angiography (gated blood pool scan), SPECT and PET over standard echocardiography or LV cineangiography. In the setting of established ischaemic cardiomyopathy, well-established PET or MRI techniques can also be used to distinguish hibernating myocardium from non-viable, post-infarct scar in order to shed light on the underlying cause of ischaemic dysfunction. This in turn may influence the choice of cell type(s) to be administered (e.g. pro-angiogenic BM or blood-derived cells for hibernation), as well as providing useful insight into the realistic likelihood of a therapeutic response.

Fig. 3.

Integrated imaging of the cell therapy patient. This flow chart depicts the extensive and multifaceted applications for imaging in cardiac patients who receive cell therapy, spanning baseline evaluation and patient selection, the delivery procedure itself and the short- and long-term surveillance of cell fate, tissue substrate response and ultimately patient outcomes. Applicable imaging strategies: 1 MRI, SPECT, PET, echocardiography, left ventriculography for myocardial structure and function; coronary angiography and flow studies, myocardial stress imaging with SPECT, MRI or echocardiography for perfusion; late gadolinium enhancement MRI, PET for viability. 2 Biplanar X-ray fluoroscopy, electromechanical mapping, MRI fluoroscopy for guided cell delivery. 3 SPECT (e.g. ¹¹¹In), PET (e.g. ¹⁸FFDG, ⁶⁴Cu), MRI (e.g. SPIO) for short-term cell fate. 4 Reporter gene strategies and SPECT (e.g. sodium-iodide symporter), PET (e.g. HSV-tk), MRI (e.g. ferritin) for long-term cell fate. 5 Molecular imaging of cellular and myocardial targets (e.g. inflammation, apoptosis, angiogenesis). Dotted lines indicate that the imaging strategy has not yet been used clinically for its nominated purpose

With respect to follow-up after cell administration, different modalities can be combined to provide complementary information in the same subject to monitor important surrogate endpoints, including global LV volumes and ejection fraction, infarct size and thickness, myocardial perfusion, viability and metabolism. Such measures should be evaluated and quantified with the most accurate imaging techniques available using standardised protocols for data acquisition and analysis, by experts in the field [139]. However, rather than adopting a blanket approach to the assessment of surrogate parameters, the objective of imaging should remain focused on capturing the most meaningful endpoints with the most appropriate tools. Surveillance of regional contractile and diastolic function, infarct size and fibrosis burden may be especially prudent as these parameters can uncover early or subtle benefits from therapy before there is global improvement in cardiac performance [140, 141]. Other measurable indices may also emerge as having important implications for cell engraftment, such as microvascular obstruction, myocardial blood flow, haemorrhage, oedema, area at risk and oxygenation [142, 143].

Furthermore, one can also envisage the future utility of molecular imaging to assess specific pathophysiological substrates and unmask valuable information about the actions of cell transfer on key targets such as myocardial inflammation, cardiomyocyte apoptosis and angiogenesis [144]. Although not yet studied in the cell therapy setting, inflammation can be monitored after preclinical MI with optical tomography to detect fluorescent signal that is released when protease-activatable probes are recognised and cleaved by specific pro-inflammatory enzymes (e.g. matrix metalloproteinases, cathepsins) [145]. Cardiac cell apoptosis has also been assessed in human patients with acute MI [146], heart failure [147] and transplant rejection [148] through SPECT imaging of technetium-labelled annexin V, an endogenous protein which targets phosphatidylserine on apoptotic cell membranes. Preclinical feasibility has also been shown for MRI detection of annexin V conjugated to magnetofluorescent nanoparticles [149]. Similarly, myocardial angiogenesis can be monitored using several molecular imaging targets such as the α_vβ₃ integrin heterodimer or vascular endothelial growth factor (VEGF) receptors, which have been targeted with SPECT and PET probes in experimental and human MI [150–153].

The inclusion of noninvasive cell tracking in future clinical trials and mechanistic assessment of stem cell actions on the myocardium will provide direct insights not only into the mechanisms by which cell therapy strategies are successful, but also where they need to be optimised. Information about cell fate both in the short and long term will help to refocus efforts on refining the mechanics of cellular delivery and/or the functional biology of different stem cell candidates. Potentially, it will offer a means to identify likely non-responders (initial failure of cell delivery or rapid loss of cell engraftment), in whom repeat treatments may be warranted, or those patients who require close monitoring of non-cardiac toxicity (distribution of cells to other organs). For example, in the specific case of intramyocardial delivery, assessment of cell retention could convey immediate feedback regarding the success and precise location of injections. During longer follow-up, this information could then be used to help interpret regional changes in myocardial function, viability and perfusion.

At this stage, the path to human application requires further efforts to ensure that tracking techniques are closely tailored to answer specific scientific and clinical questions, while satisfying the strict standards of regulatory bodies, such as the FDA and European Medicines Agency. In all cases, the safety and efficiency of specific labelling and imaging techniques should be confirmed when applied to different cell types and disease settings. As a minimum, potential cytotoxicity needs careful assessment in terms of the viability, proliferative capacity, phenotype and intended function (e.g. angiogenesis, cardiomyocyte transdifferentiation) of the administered cells. Ideally, this evaluation should be extended over days to weeks after the initial incubation with the labelling agent, to exclude delayed adverse effects [102]. Considering that much of the proposed benefit of cell therapy appears to be mediated by paracrine mechanisms [154], future studies are also advised to investigate the effects of cell labelling on cytokine and growth factor synthesis.

For direct cell imaging, many of the labelling agents are already approved for human use, albeit for different purposes, and precedents exist for short-term radionuclide tracking of cell therapy in patients with IHD. Thus, at the present time, direct labelling appears more clinically feasible from the regulatory standpoint. Longer-term cell surveillance currently falls into the experimental realm of reporter gene labelling, which faces considerable challenges before clearance by regulatory bodies (e.g. Recombinant DNA Advisory Committee of the NIH and the FDA) will allow its application in humans. However, important early progress has been made by establishing the feasibility of indirect cell imaging in large animal studies, together with limited clinical experience [128]. Commercial interest and large-scale investment in the research, development and production of new labelling agents and imaging technologies will also play an important role in accelerating the transition of cell imaging to clinical practice.

Conclusion

The challenges created by the mechanistic uncertainties after cell transfer to the diseased heart have helped stimulate rapid progress in the fields of cell labelling and noninvasive tracking. Imaging has helped to elucidate several fundamental aspects of cell therapy, including the shortcomings of cell biology and current delivery methods as they relate to in vivo cell retention and engraftment. However, at present, no single labelling or imaging strategy fulfils all of the required need and further work is required to progress cell imaging effectively and safely to clinical trials and individual patient management. In tight conjunction with the indispensable roles that imaging already plays to assist cell delivery and patient follow-up, it is easy to envisage how a multimodal approach to cell tracking will provide great additive value to the future practice of cardiovascular cell therapy.