Cell tracking in cardiac repair: what to image and how to image.

Stem cell therapies hold the great promise and interest for cardiac regeneration among scientists, clinicians and patients. However, advancement and distillation of a standard treatment regimen are not yet finalised. Into this breach step recent developments in the imaging biosciences. Thus far, these technical and protocol refinements have played a critical role not only in the evaluation of the recovery of cardiac function but also in providing important insights into the mechanism of action of stem cells. Molecular imaging, in its many forms, has rapidly become a necessary tool for the validation and optimisation of stem cell engrafting strategies in preclinical studies. These include a suite of radionuclide, magnetic resonance and optical imaging strategies to evaluate non-invasively the fate of transplanted cells. In this review, we highlight the state-of-the-art of the various imaging techniques for cardiac stem cell presenting the strengths and limitations of each approach, with a particular focus on clinical applicability.

Introduction

In the last decade a great amount of research and clinical interest has been directed at stem cells (SC) for their potential to regenerate otherwise permanently damaged tissues. Work with these pluripotent cells has begun to be broadly explored, giving new hope for regenerative approaches in the therapy of myocardial infarction (MI).

Early success in preclinical studies demonstrated that stem cell-based therapy holds the potential to limit the functional degradation of cardiac function after MI [1]. This instigated clinical translation at a rapid pace (Table 1). Since the first study in 2002 which showed safety and effectiveness on intracoronary transplantation of autologous SC [2], several randomised, controlled clinical trials have been performed. Due to the absence of standardised protocols (cell number, timing and route of injection, baseline patient characteristics and techniques of evaluating cardiac function), results have been mixed. However, recent meta-analyses have shown that improvement in ejection fraction (EF), ventricular dimension and infarct area, despite being modest, are statistically significant [3-5].

Table 1

Selected randomised clinical trials (>50 patients) of stem cell transplantation following myocardial infarction

Study	Pts	Cell type	Assessment method	Outcome
REPAIR-AMI [101, 102]	204	Intracoronary BMC vs placebo	LV angiography	At 4 months LVEF increased in BMC vs placebo (mean±SD) increase, (5.5 ± 7.3% vs. 3.0 ± 6.5%; P = 0.01). At 12 months: death, recurrence of myocardial infarction, rehospitalization for heart failure significantly reduced.
ASTAMI [103, 104]	100	Intracoronary BMC vs control	99mTc-SPECT; echo; MRI	No effect on global left ventricular function at 6 months and 3 years.
BOOST[105, 106]	60	Intracoronary BMC vs control	MRI	At 6 months global LVEF increase (6.7%). No effects at 18 months and 5 years.
Janssens et al. [107]	67	Intracoronary BMC vs placebo	MRI; [11C] acetate PET	At 4 months no effect on LVEF and LV volumes. Reduction of infarct volume (measured by serial contrast-enhanced MRI) was greater in BMC patients than in controls.
TOPCARE-AMI [16]	59	Intracoronary BMC vs CPC	LV angiography; MRI	At 4 months LV angiography showed significant increase of LVEF (50 ± 10% to 58 ± 10%), and significant decrease of end-systolic volumes (54 ± 19 ml to 44 ± 20 ml) without differences between the two cell groups. At 12 months MRI showed reduced infarct size and absence of reactive hypertrophy.
Meluzin et al. [108]	60	Intracoronary BMC (high and low doser) vs control	Echo; 99mTc-SPECT; 18F-FDG PET	LVEF improved in the group receiving the highest dose (108 cells) by 6%, 7%, and 7% at months 3, 6, and 12, respectively.
MAGIC [8]	97	SMB vs placebo injected in and around the scar	Echo	No improvement in regional or global LV function at 6 months.
Chen et al. [109]	69	Intracoronary BMSC (bone marrow mesenchymal stem cells) vs placebo	18F-FDG; Echo	At 3 months LVEF significantly increased in the BMSC group (67 ± 11%) compared to controls (53 ± 8%) and the same group before implantation (49 ± 9%). No change in LVEF at 6 months versus 3 months.
Dill T et al. [110]	204	Intracoronary BMC vs placebo	MRI	In the BMC group, EF increased significantly by 3.2 ± 1.3 absolute percentage points at 4 months, and this increase was sustained at 12 months (+3.4 ± 1.3 absolute percentage points vs baseline). In the placebo group, EF was unchanged (+0.6 ± 1.2 absolute percentage points, at 12 months.

BMC, bone marrow stem cells; CPC, circulating progenitor cells; SMB, skeletal myoblast. Pts, number of patients. LVEF, left ventricular ejection fraction

This field clearly benefited from the advancements in imaging sciences as almost all clinical trials involved the use of one or more imaging techniques to evaluate the therapeutic efficacy of stem cell transplantion. Clinically established techniques allow for the evaluation of myocardial contractility, viability and perfusion, but do not provide the direct visualisation of transplanted stem cells, therefore their effective presence and viability can be only presumed. Ideally, transplanted cells in the infarcted myocardium are expected to survive engraftment, be self-renewing and differentiate into cardiac cells (cardiomyocytes, endothelial cells or smooth muscle cells) forming electromechanical junctions with adjacent viable tissues. However, the long-term improvement appears to be most closely related to paracrine effects rather than transdifferentiation of the cell transplant and heart muscle regeneration [6].

Great strides in imaging techniques and technologies have been made that enable the cellular and molecular imaging of transplanted stem cells, their short and long term fate and in some instances their viability and differentiation status.

Stem cells for cardiac repair

Currently adult stem cells, embryonic stem cells (ESC) and induced pluripotent stem cells (iPS) can be used to regenerate heart tissue. Adult stem cells comprise skeletal myoblasts (SM), mesenchymal stem cells (MSC), bone marrow- derived stem cells (BMC), endothelial (EPC) and cardiac progenitor (CPC) cells. SM were the first option to be used in stem cell transplant as they are available from an autologous source (therefore lacking ethical or immunogenicity issues), and have been demonstrated to provide functional benefit after myocardial infarction in animals [7]. However, in a recent clinical trial no sustained benefit in the global EF was observed and increased number of early postoperative arrhythmic events was reported [8].

BMC transplantation has been shown to improve heart function in animal models [1, 9]. However, others have identified that most of the cells injected adopted a mature haematopoietic transformation and only a small number of cardiomyocytes expressed the genetic markers of the transplanted cells [10, 11]. Mesenchymal stem cells (MSC) constitute the stromal compartment of bone marrow and, importantly, are not hematopoietic. These are able to differentiate into a variety of cell types [12] and improvement in whole heart function has been described in a swine model of myocardial infarction [13]. Visceral and subcutaneous adipose tissue have been shown to contain vascular/adipocyte progenitor cells and adult multipotent mesenchymal cells (adipose tissue-derived stromal cells [14]). ASC have been reported to improve left ventricular function in animal models of myocardial infarction [15]. The circulating endothelial progenitor cells (EPC) represent a more accessible source of autologous SC and have been used in clinical trials [16]. The existence of a subpopulation of resident cardiac SC (RCSC) has been reported that is self-renewing, clonogenic and multipotent, capable of differentiating in myocytes, smooth muscle and endothelial cells [17]. Promising results have been reported in preclinical studies [18], and results of phase I clinical trials, started in 2009, are awaited with interest.

To date, ESC-derived cardiomyocytes [19] and ESC-derived endothelial cells [20] have been successfully used to treat heart disease in animal models. However several problems are related to their use, including immunological incompatibility with the host [21], the tendency to form teratomas [22] and ethical controversies. Several studies have been performed to manipulate the expression of transcription factors with the goal of transforming somatic cells (derived from an autologous source, such as keratinocytes and fat stromal cells) into induced pluripotent stem cells (iPS) [23]. These cells possess the same advantages as ESC, without the associated immunological and ethical complications. Cardiomyocytes have been successfully obtained from iPS in vitro [24] and their transplantation in animal models of infarction resulted in improved myocardial function [25].

To summarise, ESC and iPS have the greater potential for cardiomyogenesis, while the formation of new cardiomyocytes by transdifferentiation of SM and BMC has so far not been supported with convincing evidence. It should be noted that several studies have reported moderate improvements in whole cardiac function after transplantation of SM and BMC [7, 26]. It has been demonstrated that SCs are responsible for paracrine effects, consisting of the release of various cytokines or growth factors (eg. VEGF, bFGF) that increase collateral perfusion and neoangiogenesis and influence the contractile characteristics of chronically failing myocardium [26].

Imaging of stem cells

The ability to image and monitor the biodistribution, viability and possibly the differentiation status of implanted SCs is of massive clinical and research benefit. All of the pre-clinical and clinical imaging techniques have been leveraged towards this goal; each providing unique advantages and limitations. Figure 1 illustrates the major paradigms for the labelling of SC for detection by the various imaging approaches. Table 2 reports the most important preclinical studies in the field. Table 3 summarises the most relevant features of each imaging technique.

Fig. 1

Schematic representation of the current technologies available for stem cell (SC) tracking. Before implantation SC can be passively loaded with: a superparamagnetic nanoparticles that allow for the MR detection of labelled cells as areas of signal loss; b radiolabelled PET or SPECT probes. c Reporter gene approaches consist of the introduction through viral or non-viral-vectors of a reporter gene driven by a constitutive or inducible promoter. The reporter gene undergoes transcription to mRNA, which is translated into a protein that can be: 1) an enzyme (as HSV1-tk or luciferase), 2) a receptor (as transferrin receptor or hSSTR [human somatostatin receptor]) 3) a transporter (hNIS [human sodium iodide symporter]) 4) intracellular iron storage protein (ferritin). When a complementary reporter probe is administered, it concentrates or activates only at the site where the reporter gene is expressed. The level of probe accumulation is proportional to the level of reporter gene expression and can be monitored to evaluate the number of cells or the induction of a specific reporter gene

Table 2

Selected cell tracking studies of SC

Study	Species	Cell type	Detection method	Delivery	Results
MRI
Kraitchman et al. [111]	swine	MSC	SPIO	Intramyocardial -percutaneous	Detection of tranplanted cells (25.8%) up to 3 weeks.
Amado et al. [112]	swine	MSC	SPIO	Intramyocardial- percutaneous	Gradual loss of intensity of the SPIO label but retection of tranplanted cells (42.4%±15) at 8 weeks.
Stuckey et al. [113]	rat	BMC	GFP-SPIO	Intramyocardial –direct	No improvement in LEVF. Detection of tranplanted cells up to 16 weeks confirmed by MR and immunofluorescence.
Amsalem et al. [43]	rat	MSC	SPIO	Intramyocardial –direct	At 4 weeks after injection, most of the transplanted labelled MSCs did not survive and their iron content was engulfed by resident macrophages. Injection of labelled or unlabelled cells attenuate ventricular dilatation and dysfunction after MI.
Ebert et al. [114]	mice	mESC	SPIO	Intramyocardial -direct	Detection up to 4 weeks by MRI. LVEF identical between the tranplanted group and control.
Terrovitis et al. [45]	rat	hCDCrCDC	SPIO	Intramyocardial -direct	Signal void persisted after 3 weeks in both syngeneic and xenogeneic cell implantation. Immunohistochemistry identifies the iron containing cells as macrophages.
Radionuclide
Chin et al. [62]	swine	MSC	111In-oxine	Intravenous	Significant lung activity that obscured the assessment of myocardial cell tracking.
Brenner et al.[63]	rat	HPC-CD34+	111In-oxine	Intracavitary (left ventriculum)	Impairment of cell proliferation and differentiation induced by 111In-oxine. At 96 h only 1% of radioactivity was detected in the heart.
Blackwood [65]	dog	BMC	111In-tropolone	Intramyocardial -direct	Viability at day 6 after intramyocardial injection was calculated to be 75%.
Terrovitis et a.. [82]	rat	rCDC	18F-FDG	Intramyocardial -direct	Different retention values were observed at 1 h after injection of cells with normal condition (17.8%±7.3), arrested heart (75.8%±18.3), adenosine injection (35.4%±5.3) and adenosine plus fibrin glue (39.3%±11.6).
Mitchell et al. [66]	dog	EPC	111In-tropolone	Intramyocardial -percutaneous	15 days after intramyocardial injection SPECT/CT imaging demonstrated comparable degrees of retention: 57%±15 for the subepicardial injections and 54%±26 for the subendocardial injections.
Multimodal
Terrovitis et al..	rat	rCDC	99mTc, 124I; hNIS	Intramyocardial -direct	Detection up to 6 days after injection and their presence validated by ex vivo imaging and qPCR.
Qiao et al. [98]	rat	mESC	SPIO; HSV1-tk+ 18F-FHBG	Intramyocardial -direct	Increasing 18 F -FHBG uptake up to 4 weeks. Most of the SPIO were contained in infiltrating macrophages at week 4. Teratoma formation. Increased LVEF. Only <0.5% of the implanted cell were cardiomyocytes.
Chapon et al. [115]	rat	rBMC	SPIO; 18F-FDG	Intramyocardial -direct	MRI detection of SPIO labelled cells grafted in the heart up to 6 weeks, confirmed by hystology. At 1 week increased 18 F-FDG uptake in BMC implanted heart vs control. No improvement of heart function.
Higuchi et al. [99]	rat	hEPC	SPIO; NIS +124I	Intramyocardial -direct	Rapid decrease of 124I uptake after day 3. Signal not detectable at day 7. MRI signal void remained unchanged throughout the follow-up period. Histology confirmed the presence of transplanted cells on day 1 but not on day 7, when iron was contained only in resident macrophages.
Li et al. [116]	rat	RCSC	Fluc + D-Luciferin; 18F-FDG PET; echocardiography; MRI	Intramyocardial -direct	Implanted cells detected up to 7 weeks by bioluminescence. No improvement in cardiac function assessed by 18F-FDG PET, MRI, echocardiogram and invasive hemodynamic pressure volume-analysis.

BMC (bone marrow derived Stem Cells); MSC mesenchymal stem cells; mESC (mouse embryonic stem cells); hCDC (human cardiac derived stem cells), rCDC (rat cardiac derived stem cells); HPC (hematopoietic progenitor cells); hEPC, human Endothelial Progenitor Cells; RCSC (resident cardiac stem cells); NIS (sodium iodide symporter); Fluc (firefly luciferase); hNIS (human sodium-iodide symporter; MI, Myocardial infarction

Table 3

What to image and how to image

Imaging modality	Spatial resolution (mm)	Sensitivity (mol/L)	Cell Manipulation	What to image	How to image	Advantages	Disadvantages
Fluorescence Imaging	FRI: 2–3 mm; FMT: 1 mm	10−9–10−12	Cells labeled with near-infrared probes (fluorochromes, Quantum dots, etc.)	Residence, homing, quantification (FMT)	Direct imaging; at NIR wavelenghts can image deep tissue	Multiplexed imaging	Not suitable for clinical translation; relatively low spatial resolution
Bioluminescence Imaging	3–5	10−15–10−17	Cells transduced to express luciferase	Residence, homing, viability, differentiation, quantification	After systemic injection of D-Luciferine or Coelenterazine	Easy, high sensitivity, high-throughput, low cost; assessment of cell viability	Not suitable for clinical translation; surface imaging; relatively low spatial resolution; requires completely dark environment
PET	1-2 (μPET); 6–10 (clinical PET)	10−11–10−12	Cells loaded with 18F-FDG; 64Cu labelled compounds	Residence, homing, quantification	Direct imaging	High sensitivity, translational	Radiation; only short term cell tracking
			Cells transduced to express PET reporter genes (HSV1tk, HSV1-sr39tk )	Residence, homing, differentiation, quantification,	After systemic injection of correspondent radiolabelled probe (18F-FHBG, 18F-FEAU, etc.)	High sensitivity, long term cell tracking; assessment viability	Radiation; need to transduce cells; potential immunogenicity
SPECT	0.5-2 (μSPECT); 7–15 (clinical SPECT)	10−10–10−11	Cells labelled with 99mTc-, 111In-labelled compounds	Residence, homing, quantification	Direct imaging	High sensitivity, translational; assessment viability	Radiation; only short term cell tracking
			Cells transduced to express reporter genes (hNIS )	Residence, homing, viability, differentiation, quantification,	After systemic injection of correspondent radiolabeled probe ( 99mTc, etc.)	High sensitivity, long term cell tracking	Radiation; need to transduce cells; potential immunogenicity
MRI	0.01–0.1 (small animal); 0.5–1.5 (clinical)	10−3–10−5	Cells labeled with Iron Oxides; Gd or Mn chelates; perfluorocarbon (19F)	Residence, homing, migration, quantification	Direct imaging	High spatial resolution; high soft tissue contrast; functional imaging	Relatively low sensitivity; long scanning times; probe dilution upon cell proliferation; persistence of SPIO after cell death (macrophage)
			Cells transduced to express MRI reporter genes β-galactosidase, transferrin receptor, ferritin, MagA and lysine-rich proteins	Residence, homing, quantification, migration, differentiation	Direct imaging or after injection of iron oxides (transferrin receptor, ferritin)	High spatial resolution; high soft tissue contrast; functional imaging; no probe dilution;	Low sensitivity; need to transduce cells; potential immunogenicity

NIR, Near-Infra red imaging; FRI, Fluorescence reflectance imaiging; FMT, Fluorescence molecular tomography

MRI

Magnetic resonance imaging (MRI) is a widely established technique for the evaluation of cardiac anatomy and function, often through the addition of paramagnetic contrast material [27]. Taking advantage of its excellent spatial (10–100 μm [small animal MRI]); 500–1500 μm [clinical]) and temporal resolution SC labelled with superparamagnetic and paramagnetic agents can be visualised [28, 29]. Multispectral non-1H MR imaging (specifically 19F) has also been exploited to enable tracking of transplanted cells.

Superparamagnetic iron oxide nanoparticles

Superparamagnetic iron oxide (SPIO) nanoparticles provide labelled cells with a large magnetic moment and are detectable by MR imaging devices or benchtop relaxometers. SPIO functions by acting as magnetic inhomogeneities, locally disturbing the magnetic field. This leads to enhanced dephasing of protons, resulting in decreased signal intensity on T2-weighted and T2*-weighted images (Figs. 2 and 3). These nanoparticles often consist of a core of iron oxide (magnetite and/or maghemite) with a polymeric or polysaccharide coating. They are widely viewed to be biocompatible, have a limited effect on cell function and can be synthesized to be biodegradable. According to their size (diameter), these are classified as ultrasmall paramagnetic iron oxide (USPIO, <10 nm), monocrystalline iron oxide particles (MION, or cross-linked CLIO; 10–30 nm), standard superparamagnetic iron oxide (SPIO; 60–150 nm) and micron-sized iron oxide particles (MPIO, 0.7–1.6 μm). Of note, ferucarbotran (Resovist®; Bayer Schering Pharma, Berlin, Germany) and ferumoxides (Feridex I.V.®, Advanced Magnetic Industries, Cambridge, Maryland, USA; Endorem®, Guerbet, Gorinchem, the Netherlands) have been approved by the FDA for contrast enhanced-MRI imaging of liver tumors [30] and metastatic involvement of lymph nodes [31].

Fig. 2

Anatomical and functional MR evaluation after transplantion of adipose-derived stem cell (ASC) and relative controls: cell culture medium (CCM), and untreated hearts. The CCM-treated and untreated hearts showed evident thinning in the anterior wall of the left ventricle. From Wang, L. et al. Am J Physiol Heart Circ Physiol 297: H1020-H1031 2009 [15] (with permission)

Fig. 3

Longitudinal BLI and MRI of H9c2 cells after transplantation. BLI shows a robust distinct heart signal on day 1 (red arrow), compared to no discernable signal in a representative control rat having received non-labelled cells (top panel, left). The signal increases slightly on day 3 but decreases rapidly to near background levels by day 6. MRI imaging of a representative rat injected with the same amount of cells labelled with Feridex shows a large hypointense signal (red arrow) in the anterolateral wall of the myocardium. The size of the signal decreases slightly over time, and the signal persists for at least 80 days post cell injection. No signal is observed in control rat that received non-labelled cells (bottom panel, left) Chen, I. Y et al. Mol Imaging Biol. 2009 May-Jun;11:178–87. [42] (with permission)

Cell uptake is mediated through the size and electrostatic charge conditions of the SPIO [32], schematically illustrated in Fig. 1a. Further, loading can be augmented through the addition of cell penetrating peptides, electroporation or transfection agents [33].

Studies reported that SPIOs do not affect cell viability, proliferation, differentiation or migration [34-38]. However, recent work has raised several concerns, such as decreased MSC migration and colony-formation ability [39], and interference with cell function [40, 41]. A major issue beyond potential cellular effects is the question of contrast specificity to the presence of cells. Namely, the hypointense signal is maintained at a site regardless of cell viability and SPIO are present not necessarily within implanted SC at longer time points [42], but rather in phagocytosing monocytes following SC death [43]. Recently, Winter et al. reported the absence of any discrimination between healthy successfully engrafted SC and dead SC phagocytosed by macrophages within the heart. In particular, no differences in signal voids up to more than 40 days were observed with dead and viable cells recipient with respect to size, number and localisation [44]. Similarly, it has been demonstrated that MRI overestimates the SPIO labelled SC survival after transplantation in the heart [45]. Furthermore, SPIO-induced hypointensity can sometimes be difficult to interpret because it may be obscured by the presence of endogenous blood derivates, such as hemosiderin [46]. The clinical translation of SPIO for cell tracking is further reduced now that ferumoxides (Feridex® or Endorem®) are no longer available in the USA and Europe. However, the use of iron oxides approaches should not be discouraged as they provide very high sensitivities. New compounds with improved tissue clearance properties (therefore higher specificities) are awaited from material sciences research.

Paramagnetic ions

Cell labelling with “positive contrast” such as gadolinium (Gd) chelates and manganese (Mn) chloride compounds allow the visualisation of SC as hyperintense signal on T1-weighted images. Internalization of Gd can be accomplished by exposure of cells to Gd chelates or through the use of liposomal formulations [33]. MRI sensitivity in the detection of Gd-labelled cells is lower compared to SPIO-labelled cells and is dependent upon contrast behaviour and relaxivity in the cellular compartment (endosomes) in which they are localised [47, 48]. This is a result of the reduced water accessibility to chelated ions following intracellular concentration, resulting in decreased relaxivity. To overcome these issues several approaches have been considered to drive the endosomal escape of paramagnetic compounds [49]. Furthermore, safety issues might be related to the rapid dechelation of compounds at the low pH of lysosomes and endosomes raising concerns related to free Gd3+ ions [50].

Sub-millimolar concentrations of Mn chloride (MnCl2) have been sufficient to enable SC labelling and detection for both in vitro and in vivo MR, with no detectable toxicity. Also in the same study the potential of MnCl2 labelling in the assessment of SC viability by T1 and T2 mapping was investigated in in vitro studies [51]. Mn-oxide nanoparticles have recently been used to label and track implanted glioma cells. Of considerable interest, the feasibility of successfully tracking two cell populations simultaneously has been suggested, where one is labelled with Mn-Oxide and the other with SPIOs [52]. These and other paramagnetic ion techniques offer the hope that positive contrast approaches will enable sensitive MR tracking of SC in vivo. Novel nanotechnology approaches are becoming available for stem cell tracking such as gadolinium-containing carbon nanocapsules (Gadonanotubes), whose T1 relaxivity is greater than that of any known material to date (outperforming clinically available Gd-based contrast agents by 40-fold) [53]. They will definitely play a role in the future of imaging sciences as soon as their toxicity profiles, currently under investigation, have been clarified.

19F MR

Similar to the imaging of relaxation of 1H from water, 19F can be used as the basis of the signal for MR spectroscopy and image formation. While this technique is not implemented widely in the clinic, there are unique advantages to fluorine-MR that make it an attractive option for SC tracking in myocardial applications. 19F is not present naturally in soft tissues therefore its signal is exclusively derived from the exogenous contrast agent applied, be it a perfluorocarbon particle or fluorinated nucleosides [54]. 19F MRI can be used with existing 1H imaging hardware since 19F and 1H gyromagnetic ratios differ by only 6%.

Importantly, 19F signal can be overlaid on 1H-MR anatomical images for a highly selective, high-resolution map of cell transplantation. This technique allows for quantitative determination of the cell population [55]. A perfluorocarbon particle loaded cell scheme has been used to show the unequivocal and unique signature for SC, enabling spatial cell localization via19F- MRI and quantitation via 19F-spectroscopy [56]. Perfluorocarbons have been extensively studied and used as blood substitutes, therefore their toxicity profiles are known. 19F cell tracking has attracted interest, but is still at an early stage of development. It should be noted that this method does suffer from the drawback of lower sensitivity requiring longer imaging times. Efforts are underway to address these deficits including imaging hardware, imaging sequences, and label improvement and 19F MR imaging is expected to play a role in cell tracking in the future [54].

Radionuclide imaging

Imaging of SC has also taken advantage of the high sensitivity (10-10−10−12 mol/L vs 10−3–10−5 mol/L of MRI) and quantitative (acute cell retention as a percentage of the net injected dose per weight, [%ID/g]) characteristics of radionuclide imaging [57]. However, PET and particularly SPECT have inferior spatial resolution (1–2 mm) compared with MRI. Moreover, radionuclide-labelled cells can only be visualised as long as the radioactivity is still detectable (e.g. 18F: 110 min; 111In: 2.8 days; 99mTc: 6 h). This sets an appreciable limitation on the radioisotopes direct labelling value for medium- and long-term SC transplant monitoring. SPECT has been largely used to investigate the short-term fate of transplanted cells labelled with radioactive compounds such as 111In-oxine [58-63], 99mTc-hexamethylprophylene amine oxine (HMPAO) [64] or 111In-tropolone [65, 66]. A persistent limitation for deployment of SPECT is that in order to generate useful (quantitative) images within a reasonable time frame, the administration of relatively large doses of radioactivity are required. This poses the concern of inherent radiation damage (reduced viability and proliferation). In the case of 111In, Auger electrons are also emitted leading to adverse biological effects in very short distances (from the nm to μm range). Brenner et al., demonstrated that despite the homing of progenitor cells into the infarction area, cell labelling with 111In-oxine impairs significantly the viability, proliferation and differentiation at 48 h after implantation [63]. Similar results were observed after exposure of murine haematopoietic progenitor cells at even much lower levels of radioactivity [67]. The use of other compounds, such as 111In-tropolone, inhibited cell proliferation 3 days after labeling [68]. To abrogate these effects, it has been suggested that only a fraction of the SC population be labelled [69]. Regardless of the method used, very few studies have reported the absence of any cell function impairment [58, 62]. These studies underline the need for further in vitro studies considering different SC, exposed to different activities and importantly following the same labelling protocol.

Positron emission tomography (PET) has been regarded as having higher sensitivity (2 to 3 orders of magnitude) and better spatial and temporal resolution than SPECT [70]. 18F-Fluorodeoxyglucose (18F-FDG) has been used for cell labelling and short term imaging in preclinical [71] and clinical settings (Fig. 4) [72, 73]. After intracoronary injection all stem cells showed poor engraftment regardless of cell type and number of implanted cells [61, 72, 73]. In all cases intravenous injection of SC did not show detectable homing of cells to the myocardium [72, 73].

Fig. 4

PET/CT images of a patient with hystory of anterior wall infarction. After percutaneous intervention 18 F-FDG labelled cells were implanted via intracoronary catheter and images obtained at 2 hrs after the procedure. Total amount of SC at the injection site was measured (2.1% of injected dose). From Kang et al. J Nucl Med 2006; 47:1295–1301. [73] (with permission)

Augmenting the higher sensitivity, the wider availability of hybrid PET-CT systems allows for a combination of anatomical non-invasive coronary angiography and cell tracking. This multimodal imaging capability and clinical availability are tempered somewhat by the the short half life of 18F. Isotopes with longer half life, such as 64Cu (12.7 h) have been suggested [74]. However, with radionuclide based techniques pursued so far, only the immediate fate of transplanted stem cells can be interrogated.

Reporter genes

Reporter gene approaches have significant potential to reveal insights into the mechanisms and fate of SC therapies. The reporter gene paradigm requires often the appropriate combination of reporter transgene and a reporter probe, such that the reporter gene product has to interact with an imaging probe (optical, nuclear, magnetic) and when this event occurs the signal may be detected and quantified with the corresponding imaging technique (Fig. 1c).

Several advantages of reporter gene approaches have been described [75]. Namely, this system identifies with exquisite specificity only viable cells (which actively contain the gene product) and allows long term tracking of transduced SC (circumventing issues of probe dilution with cellular proliferation). Reporter genes can be designed as “constitutive” whose signal is “always turned on” (suitable for the evaluation of transplantation, migration and proliferation of stably transduced SC) or “inducible” reporter gene which is activated and regulated by specific endogenous transcription factors and promoters [75, 76] providing a non-invasive readout of information regarding SC differentiation.

The most widely used reporter gene for radiotracer based imaging is HSV1-tk (Herpes simplex virus type 1 thymidine kinase) and its mutant, the HSV1-sr39tk. Unlike mammalian TK1, this enzyme efficiently phosphorylates purine and pyrimidine analogues which results in trapping and accumulation of these ligands. It has been successfully used in association with 18F or 124I -2′-deoxy-2′-fluoro-5-iodo-1-[β]-D-arabinofuranosyluracil (18F-FIAU and 124I-FIAU), 18F 2′-fluoro-5-ethyl-1-[β]-D-arabinofuranosyluracil (18F-FEAU), and 9-(4-[18F]fluoro-3-hydroxymethylbutyl)guanine (18F-FHBG) [75, 77].

Wu et al., pioneered the reporter gene approach in the heart by imaging in vivo transplanted cells (expressing luciferase or HSV1-sr39tk) up to 2 weeks by 18F-FHBG PET imaging or BLI [78]. Furthermore, Cao et al., reported survival and proliferation (through increasing signal up to 4 weeks) of murine ES stably transduced with a triple fusion reporter gene, enabling simultaneous PET, bioluminescence and fluorescence imaging [79].

Despite the advantage of signal amplification (through probe phosphorylation and accumulation within cells) of HSV-tk based approaches, its immunogenicity might limit use in humans [80]. To overcome this limitation, the human mitochondrial thymidine kinase type 2 (hTK2) have been proposed [81]. An alternative is the sodium iodide symporter [51] as a PET and SPECT reporter gene used in conjunction with 124I or 99mTc (pertechnetate), respectively [75, 82]. Here, despite the lack of probe/signal amplification observed in receptor- and transporter-based techniques (as 124I or 99mTc are free to diffuse out of the cells), hNIS is not immunogenic (since it is expressed in the thyroid, stomach salivary gland, choroid plexus but not in the heart) and does not require complex radiosynthesis of the probes. Nevertheless, in reporter gene approach for the imaging of SC-based cardiac therapy several important issues remain. First, the non-physiological expression of reporter gene proteins may perturb the critical SC cellular and therapeutic functions. To be fully reliable, this system has to guarantee the long term expression of the reporter gene in the proliferating population. Adenoviral transfection is hampered by episomal gene expression (the reporter gene is not integrated in the chromatin, and because they are not replicating, they become diluted with cell proliferation) and by immunogenicity (leakiness of immunogenic adenoviral proteins that can lead to an immune response) [83]. On the other hand, lentiviral vectors accomplish the integration of the reporter gene in the host cell chromatin allowing stable expression in dividing cells [84] and circumventing immunogenicity [85]. Even when lentiviral vectors are used however, transgene expression can be silenced by DNA methylation especially when strong promoters, such as CMV, are used to drive the expression of the reporter gene [86]. The integration of the reporter gene within the genome has raised concerns about the risk of mutagenesis and potential oncogenicity [87].

The imaging of differentiation in vivo was recently investigated by Kammili et al., by employing a novel dual-reporter mouse embryonic SC line. Here, enhanced yellow fluorescent protein (EYFP) was used as a “constitutive reporter”, and the firefly luciferase reporter as an “inducible reporter”. This latter gene was under the control of the cardiac sodium-calcium exchanger 1 (Ncx1) promoter which showed increased activity upon differentiation of SC into beating cardiomyocytes [76].

Several candidates have been proposed as MRI reporter genes such as β-galactosidase, transferrin receptor, ferritin, MagA and lysine-rich proteins [88, 89]. Recently, SM engineered to express ferritin have been transplanted in infarcted heart and detected (as decreased signal up to 25%) up to 3 weeks [90]. Several studies have been reported with the application of MR reporters, however, none of these strategies have led to a significant number of follow-up studies. This is due to the low sensitivity of MRI for imaging of gene activity in vivo.

Optical imaging

BLI

In contrast to the immediate clinical impact of magnetic and nuclear tomographic imaging, optical imaging techniques such as bioluminescence, planar and fluorescence-mediated tomography have been largely restricted to use in preclinical models. Bioluminescence imaging is commonly used for cell tracking in SC transplantation studies [78, 79, 91] (Figs. 3 and 5). SC are transduced with a luciferase gene and implanted in the recipient animal. Following injection, the probe (D-Luciferin) is oxidized only in the cells expressing luciferase in presence of ATP, O2 and Mg2+ resulting in light photons being emitted (which can be detected by ultrasensitive charge-coupled device [CCD] cameras). BLI has many advantages: it is highly sensitive, quantitative, simple and inexpensive. However, the barrier to clinical translation lies in the inherent limitations imposed by poor tissue penetration (1–2 cm) (allowing only surface imaging), high rates of scattering of visible wavelength photons on the human scale and low resolution (3–5 mm) (which hamper the exact evaluation of the exact location of the cells) [57].

Fig. 5

Bioluminescence imaging of CD34+ cells expressing the TGL gene (HSV1-tk, e-GFP, f-luc) and implanted in the heart of a SCID mouse. Systemically administered luciferin is activated (oxidized by luciferase) in the injected cells. Here we see follow-up of implanted cells up to 52 weeks post-implantation. Measurement of emitted light in CD34+ implants is higher than in controls (PBS injection). From Wang, J. et al. Circ Res 2010;106:1904–1911. [91] (with permission)

Fluorescence

Direct labelling of SC with fluorescent probes for visualisation in vitro and in vivo has been fueled by the availability of near infrared (NIR) probes, as their spectral properties are matched to lower tissue attenuation in the so-called NIR-window. This provides greater signal penetration of tissue through reduced light absorption and tissue scattering. Therefore they have clinical potential, however limited to near-surface or intraoperative imaging stem cell tracking.

Near infrared imaging provides high sensitivity as well as tomographic capabilities and there is no evidence at present that dyes released after cell death are taken up by macrophages. Intracoronary delivery of MSC labelled with the NIR dye IR-786 has been successfully tracked in a swine model of myocardial infarction and sensitivity of 10.000 cells has been reported [92].

Quantum dots (QD) are a class of inorganic, fluorescent nanoparticles that have been successfully used to label SC. Biocompatibility of QD at low concentrations has been demonstrated in vitro in MSC cultures [93] and the absence of adverse effects on cell viability, proliferation or differentiation reported [94]. One of the most attractive qualities of these nanoparticles is their capacity for multiplexed imaging. The tracking of different cell populations is concurrently achieved by labelling cells with different QDs. Multiplex optical imaging of QD-labelled embryonic stem cells have been reported up to 14 days from injection in mice [94]. Moreover, it has been shown that single QD-MSC can be detected in histological sections for at least 8 weeks after delivery [95]. The long-term effects on SC functionality are still unknown, however concerns are related to their metallic core include its exposure or dissolution which may result in toxicity, particularly from heavy metals such as Pb, Cd and Se [96, 97]

Multimodal imaging

The possibility of complementing the sensitivity of radionuclide or optical techniques with the high-resolution anatomical information from MRI is a key player in the clinical and research future of SC tracking. To date, the most interesting approach has been to develop transgenic cells that carry an optical/nuclear imaging reporter together with passive labelling with MRI contrast agents before administration. Qiao et al. assessed the survival and proliferation of SPIO-labelled murine ESC transduced with HSV1-sr39tk longitudinally (4 weeks) following injection into the healthy or infarcted myocardium. ESCs grafted and underwent proliferation, as shown by increasing uptake of 18F-FHBG in PET (Fig. 6) and decreasing the size of MR hypointense areas due to SPIO dilution. Interestingly at week 4 the majority of SPIO labels (released upon cell death) were phagocytized and contained in infiltrating macrophages rather than the ESC. Despite teratoma formation, a slight increase in left ventricular ejection fraction in ESC-treated animals was observed, mainly as a result of paracrine effects, as cardiac differentiation of implanted ESC was less than 0.5% [98]. In a similar study human EPC derived from CD34+ mononuclear cells of umbilical cord blood were transduced with NIS reporter gene and labelled with SPIOs. Rapid loss of viable grafted cells was observed, as 124I PET accumulation decreased below detection limit at 3 days after transplant. However, MRI signal void resulting from SPIO persisted, corresponding to retention of SPIOs within macrophages after graft cells’ death (Fig. 7) [99]. Triple fusion reporter gene have been widely applied for multimodality fluorescence, bioluminescence and nuclear imaging approaches [100]. Recently, in a quad-modal optical, PET, CT and MRI coregistration approach CD34+ cells were transduced with a triple fusion reporter gene (e-GFP, f-Luc and HSV1-tk). Bioluminescence imaging revealed that cells persisted in the heart up to 12 months and MRI studies reported improvement in the left ventricular ejection fraction was preserved up to 6 months (Fig. 5) [91].

Fig. 6

Co-registration of MRI a and 18 F-FHBG PET b of murine ESC transduced with HSV1-sr39tk and passively labelled with SPIO. Images depict the presence and tracking of SC 14 days after transplantation. This hybrid imaging c approach leverages the advantages of each technique; the fine anatomical resolution of MR and the specificity of nuclear imaging. From Qiao et al. Radiology 250:3, 821–829. [98] (with permission)

Fig. 7

a MRI (upper row), 124I-PET (middle row), and fusion images 13 N-NH3 (gray scale)/ 124I (colour scale) (bottom row) of rat heart 1 day after injection of EPC labelled with iron (left), NIS only (middle), or both iron and NIS (right). Signal void of iron-labeled HEPCs is observed by MRI whereas HEPCs expressing NIS showed focal 124I accumulation by PET. b Consecutive myocardial sections showing the presence of transplanted cells: autoradiography for 124I uptake mediated by NIS reporter (left), X-galactosidase staining for LacZ gene expression of graft cells (middle), and Prussian blue staining for iron particle detection (right). c Mean±SD time–activity curves after 124I administration of transplanted cell and left ventricular blood measured by PET. From Higuchi T et al. J Nucl Med, 50:1088–1094. [99] (with permission)

Conclusion

Many of the approaches to image stem cells are promising but further work is required before a wide clinical translation becomes reality. Beyond the unresolved safety and ethical issues, crucial questions: “What is the best route for cell delivery ?” “What kind and how many cells ?” and “When to inject ?” remain.

It has become clear that there is no single ‘best method’ in cell tracking. Rather there is an array of high sensitivity, high spatial resolution and functional techniques that work best in combination. The persistent trends in molecular imaging are: to focus on the development of novel MR-compatible probes able to monitor and track with sufficient sensitivity and specificity the fate of transplanted cells, new PET/SPECT reporter genes with lessened immunogenicity and oncogenicity issues, and the application of related radioprobes with better pharmacokinetic profiles.